-------
- 24 -
LIST OF TABLES
Page
Table 3-1.
Table 3-2.
Table 3-3.
Table 3-4.
Table 3-5.
Table 6-1.
Table 6-2.
Table 6-3.
Table 6-4.
Table 6-5.
Table 6-6.
Table 7-1.
Table 7-2.
Estimated Monthly Average Dry Depostion 3-15
Velocities (cm s"1) of Sulfur Dioxide at APN
Sites.
Estimated Monthly Average Dry Depostion 3-15
Velocities (cm s"1) of Particulate Sulfate
at APN Sites.
Definition of Concentration and pH Ranges for 3-17
the Trajectory Cases.
Major Source-Receptor Analyses Using Trajectories 3-32
or Storm Tracks.
A Breakdown of Natural and Anthropogenic Sulfur 3-44
Emissions (Tg.S. yr"1)
Phase II Improved United States and Canadian 6-2
SC>2 Emissions on a State and Province Basis
(Kilotonnes/Yr.) - 1980.
Phase II United States and Canadian SC>2 " 6-3
Emissions for the 63 ARMS Areas (Kilotonnes/
Yr.) - 1980.
Combined U. S. - Canadian Top 50 Sources of SC-2 6-4
Emissions - 1980.
Phase II Targeted Sensitive Areas for Work Group 6-6
2 Modeling.
Canadian Regions for Phase III Transfer 6-8
Matrices.
U. S. Regions for Phase III Transfer Matrices. 6-9
Key to 11 Source Regions and 9 Sensitive Areas. 7-3
Phase I Transfer Matrix of Annual Wet Deposition 7-5
of Sulfur (kg.S. ha."1 yr.""1) Per Unit Emission
(Tg.S.yr.-1)
-------
- 25 -
Table 7-3.
Table 7-4.
Table 7-5.
Table 7-6.
Table 8-1.
Table 8-2.
Table 8-3.
Table 8-4.
Table 8-5.
Table 8-6.
Table 8-7.
Table 8-8.
Table 8-9.
Page
Phase I Transfer Matrix of Annual Wet Deposition 7-9
of Sulfur fkg.S. ha."1 yr."1) Per Unit Emission
(Tg.S.yr.'1)
Selected Transfer Matrix Elements Values of 7-11
Annual Wet Sulfur Deposition from Phase I and
Phase II.
Significant Changes in Phase I and II RCDM and 7-13
ASTRAP Annual Wet Sulfur Depositions.
Model Estimates and Observations of Annual Wet 7-14
Sulfur Deposition (kg.S.ha."1 yr."1) at the Nine
Targeted Sensitive Areas.
AES-LRT Model Parameters Used in the Sensitivy 8-3
Study.
Seasonal Variations In The AES-LRT Model Transfer 8-4
Matrices of Absolute Values (1978).
Seasonal Variations In The AES-LRT Model Transfer 8-6
Matrices of Per Cent of Total or of Annual
Average (1978).
Seasonal Variation in Parameter Values For the 8-8
ASTRAP Model.
January and July 1978 ASTRAP Model Estimates of 8-9
SC>2 Concentrations and Wet Sulfur Depositions
at the Nine Targeted Sensitive Receptors.
Sensitivity Index - Fractional Change in Wet 8-13
Sulfur Deposition As a Function of Fractional
Change in Parameter Value Annual (din Dep/dln
Parameter).
OME-LRT Model Sensitivity of the Wet Sulfur 8-17
Deposition Factor for the Idealized Source-
Receptor Geometry Shown in Figure 8-1.
Range of Parameter Variation for the MEP-TRANS 8-21
Model Sensitivity Analysis.
Sensitivity of Wet 804 Deposition to Variations 8-23
in MEP-TRANS Model Parameters.
-------
- 26 -
Table 9-1.
Table 9-2.
Table 9-3.
Table 9-4.
Table 9-5.
Table 10-1.
Table 10- 2.
Table 10-3.
Page
NO Kinetic Rate Constants (h"1) Used in the 9-1
CAPITA Model.
Parameter Choice For 1978 MEP-TRANS Model 9-2
Simulations of NOX.
Deposition Parameters for NOX Chemistry Used in 9-3
the AES-LRT Model.
Normalized Transfer Matrix for Total Nitrogen 9-12
from the MEP-TRANS Model (kg.N. ha.-iyr."1 per
Tg. N. yr."1)
Correlations Between Observed and Predicted 9-13
N03~ From the AES-LRT Model.
Sulfate Emission Factors and Sulfur Oxide 10-5
Emission Rates.
Transfer Matrix for January 1978 Sulfur Dioxide 10-14
and Sulfate Concentrations (jug m~^) from the
ASTRAP Model Using the Phase III State/Province
S02 Emission Inventory and Primary Sulfate
Emission Factors in Table 10-1.
Transfer Matrix for July 1978 Sulfur Dioxide and 10-16
Sulfate Concentrations (/ag m"~3) from the ASTRAP
Model Using the Phase III State/Province S02
Emission Inventory and Primary Sulfate Emission
Factors in Table 10-1.
-------
-27-
LIST OF ACRONYMS
BRCG or RCG - Bilateral Research Consultation Group (U.S.-Canada)
MAP3S/RAINE - Multi-State Atmospheric Power Production Pollution
Study/Regional Acidity of Industrial Emissions
EPA - Environmental Protection Agency (U.S.)
AMS - American Meteorological Society
LRTAP - Long-Range Transport of Air Pollutants
AES - Atmospheric Environment Service (Canada)
CMC - Canadian Meteorological Centre
SURE - Sulfate Regional Experiment (EPRI)
UTM - Universal Transverse Mercator
EPRI - Electric Power Research Institute
ENAMAP - Eastern North American Model of Air Pollution
ASTRAP - Advanced Statistical Trajectory Regional Air Pollution
Model
1-D - one-dimensional
EURMAP - European Regional Model of Air Pollution
SRI formerly Stanford Research Institute, now SRI International, Inc
UNIVAC - name of a computer company
NEDS - National Emissions Data System (EPA)
OME - Ontario Ministry of the Environment
GCA - formerly Geophysics Corporation of America now GCA Corporation
NOAA/ATDL - National Oceanic and Atmospheric Administration/
Atmospheric Transport and Dispersion Laboratory
RCDM - Regional Climatological Dispersion Model
TRI - Teknekron Research, Inc.
-------
-28-
CAPITA - Center for Air Pollution Impact and Trend Analysis
MEP - Meteorological and Environmental Planning, Ltd.
TRANS - Transport of Regional Atmospheric Nitrogen and Sulfur
APN - Air and Precipitation Monitoring Network (Canada)
CANSAP - Canadian Network for Sampling Precipitation
NADP - National Atmospheric Deposition Program (U.S.)
ACID - Atmospheric Contribution to Inter-Regional Deposition
DOE - Department of Energy
ARMS - Acid Rain Mitigation Studies
RMS - Root-mean-square-error
RMSB - Standard deviation of residuals
-------
Chapter 1
INTRODUCTION
1.1 General
The Atmospheric Sciences and Analysis Work Group was
established under the Memorandum of Intent in order to provide
information, based on cooperative atmospheric modeling and
analysis of monitoring network and other data/ which would
lead to a further understanding of the transport of air pollu-
tants between source regions and sensitive areas. In addition,
the Group was to prepare proposals for the "Research, Modeling
and Monitoring" element of an agreement. The Terms of Reference
of the Group, Work Group Membership, and Glossary of Terms are
contained in Appendices 1, 2 and 3, respectively.
1.2 Phase II Activities
During Phase II the Work Group activities have been
structured in three activity areas with purposes as follows:
1. Atmospheric Sciences Review - assess the appropriate-
ness of the methods and assumptions used in regional
models to quantify source-receptor relationships;
2. Simulation Modeling - document, evaluate, inter-
compare and apply available practical regional
models; and
3. Data Analysis Review - use data to establish indepen-
dentlyfT)the usefulness of regional models and
(2) the validity of computed source-receptor
relationships.
Appendix 4 contains a listing of the agendas, attendance rosters,
and minutes of the several Modeling Subgroup workshops and
Work Group 2 meetings held during Phase II, while Appendix 5
contains a listing of Work Group 2 Phase I and II reports.
-------
1-2.
Although many substances may undergo transboundary
atmospheric transport and have harmful effects upon either
the atmosphere or surface receptors, acid deposition has been
the phenomenon of primary concern for the first two phases
of our Work Group activities. As a consequence, the highest
priority has been given to the study of oxides of sulfur and
nitrogen, the main precursors of acidic deposition. Because
of the analysis needs of the other Work Groups to develop con-
trol measures that would be effective in reducing transboundary
acid deposition effects, significant emphasis has been
placed on the development of the "transfer matrix" concept.
The transfer matrix specifies the contribution of individual
source areas to receptor areas of interest. It assumes a
linear relationship between sources and receptors, and is
generated from mathematical models of long range transport
of air pollutants. Transfer matrices have only been con-
structed, to date, for sulfur compounds for seasonal and
yearly averaging periods. Additionally, a very tentative
effort to construct an annual average nitrate transfer matrix
has been attempted.
The purpose of this Phase II report is to provide as
complete a response as is currently as possible to all the
scientific and technical areas identified in the Terms of
Reference and as specified in the approved work plan of Work
Group 2.
-------
1-3'
1.2.1 Modeling Subgroup
During Phase II/ the Modeling Subgroup of Work Group 2
has devoted its efforts to:
(1) a synthesis of the Phase I transfer matrices into a
"best estimate" for preliminary assessment iterations
(2) evaluation of selected sulfur regional transport models
against monitoring data for ambient concentrations and
deposition rates;
(3) intercomparison of model results in order to analyze their
performance characteristics and further improve their
reliability;
(4) production of the Phase II transfer matrices using
improved Phase II input data bases, and identification
and analysis of the variations and uncertainties in the
individual matrix elements;
(5) production of preliminary transfer matrices for nitrogen
oxides using several simplified NOX-N03 chemistries in
selected models;
(6) preparation for Phase III activities, including review
of the regionalization for source areas and of emissions
data bases.
A comprehensive report on the efforts of this Subgroup and
separate reports on its several activities have been prepared
(see Appendix 5).
1.2.2 Atmosheric Sciences Review and Monitoring Data
Analysis Subgroups
Many advances in understanding the regional and long-
term transport of air pollutants have been gained in recent
years, in large part due to an expansion of research efforts
on the underlying physical phenonmena involved in transport,
transformation and deposition. These efforts have driven
the development and use of large mathematical models to
-------
1-4.
integrate the scientific information as it becomes available.
Even so, it is not possible to describe fully in a practical
model all aspects of air pollution transport on a regional
or continental scale. Consequently, many simplifications
have been made in the modeling analyses presented and dis-
cussed in this report. The role of the Atmospheric Sciences
Review effort is to provide scientific analyses in selected
areas of basic atmospheric processes, so as to specify more-
precisely the validity and range of uncertainty that character-
ize the modeling methodologies utilized and presented in
this and subsequent reports. A separate report on the Phase II
efforts of the Atmospheric Science Review Subgroup has been
prepared (see Appendix 5).
Similarly, a Monitoring Data Analysis Subgroup has been
constituted to analyze and interpret available monitoring data
in order to gain further insights into transboundary air
pollution phenomena. The initial work effort of the Monitoring
Data Analysis Subgroup is presented in Chapter 3 of this
report. The Monitoring Data Analysis Subgroup (to be called
the Monitoring Interpretation Subgroup in Phase III) will
produce a separate report in Phase III.
1.3 Phase II Report
This report is structured to follow closely the terms of
reference and work plan for the Group. The following chapter
-------
1-5.
contains summaries of the four atmospheric science issues reviewed
during Phase II. Chapter 3 contains an analysis of monitoring
results and their interpretation. The next two chapters
describe the role of models in the particular application at
hand, and the status of those models which have been selected
for use in Canada and the United States. The Phase II improved
inventories for current S02 emissions in eastern Canada
and the U.S. at the state/province level and at the Acid
Rain Mitigation Study (ARMS) area level and for the top 50
S02 emitters in North America are presented in Chapter 6.
presented. The seventh chapter presents the Phase II source-
receptor transfer matrices for sulfur oxides and compares them
to those for Phase I. Chapter 8 contains a preliminary analyses
of the variations and uncertainties in the transfer matrices for
sulfur oxides.
Chapter 9 and 10 contain preliminary transfer matrices
for nitrogen oxides and primary sulfates, respectively.
Chapter 11, "Conclusions, Recommendations, and Phase III
Work Plan", charts the future course of action for the Work
Group. The final chapter contains preliminary proposals that
should be considered for inclusion within the Research, Modeling,
and Monitoring element of a transboundary agreement.
-------
1-6.
1.4 Phase III Activities
During Phase III, the Work Group will give additional
emphasis to:
(1) expanding in scope and depth the Atmospheric Science
Review and Monitoring Data Analysis efforts;
(2) exercising the selected models using the Phase II improved
SC>2 emission inventory on a state/province basis and
attempting to unify the transfer matrices;
(3) examining and quantifying the "background contributions"
to concentrations and depositions in the targeted sensitive
areas and the ventillation out of the region, particularly
to north and east;
(4) preparing the 1979 data bases for the second and third
rounds of model evaluation;
(5) exercising the selected models on additional periods of
meteorological data to assess the seasonal and annual
variabilities in source receptor-relationships;
(6) preparing and implementing a detailed work plan for modeling
additional transboundary air pollution issues; and
(7) addressing the issues raised by the peer and other external
reviews.
-------
Chapter 2
ATMOSPHERIC SCIENCE ISSUES IN LRT MODELING
2.1 Introduction
When nearing the completion of Phase I activities and
v
preparing the Phase I report, Work Group 2 held a workshop in
Washington, D.C. on December 16, 1980. A wide ranging
discussion occurred regarding the most important areas in the
atmospheric sciences which were closely related to the use of
long-range transport models. From these dicussions four
topics were chosen as requiring highest priority for immediate
review in the Phase II activities. Designated members of the
Work Group were assigned the task of preparing review papers
with the help of outside experts. These review papers were
to highlight the state of knowledge in the particular topic
area and to indicate how that knowledge is reflected in the
various models being used by the Work Group.
The four review papers are presented in full in a
background document (Report No. 2-14) by the Atmospheric
Science Subgroup of Work Group 2. A brief summary of the
most important and relevant (to modeling) findings of these
reviews follows.
2.2 Sulfur and Nitrogen Chemistry in LRT Models
Present understanding of the homogeneous gas phase
reactions of SC>2 indicates that the rate of S02 oxidation in
the atmosphere is dominated by free radical reaction processes.
-------
2-2.
The free radical species identified as important contributors
to the SC>2 oxidation process are hydroxyl (HO), methylperoxyl
(CH302) and other organic peroxyl species (RC>2/ R'C>2 etc.).
The concentration of these radicals in the atmosphere are
dependent on many factors/ the more important of which are
the concentration of volatile organic compounds and nitrogen
oxides (NO and NO2) in the atmosphere, temperature and solar
intensity. Theoretical estimates have shown that maximum S02
oxidation rates of 4.0% h~l are possible in polluted atmospheres.
However, recent experimental rate constant determinations for
the H02 and CH302 reactions with S02 indicate that these
processes may not be as important as previously thought and
that the maximum possible homogeneous S02 oxidation rate
under optimum atmospheric conditions may only be of the order
of 1.5% h"1. This rate is the result of the reaction of S02
with the hydroxyl radical only.
Present knowledge of heterogeneous pathways to S02
oxidation in the atmosphere indicates that the liquid phase
i 2
catalyzed oxidation of S02 by Mn ion and carbon are
potentially important processes, as is oxidation by hydrogen
peroxide. Theoretical estimates of the maximum rate of
atmospheric S02 oxidation via these processes are of the
order of 10% h~l. Unfortunately, a great deal of uncertainty
surrounds the actual availability of these catalyzing substances
in ambient fine particulate matter. The quantitative determination
-------
2-3
of rates of SC>2 oxidation via these processes has never been
demonstrated under actual atmospheric conditions.
Organic and nitrate particulate matter forming processes
are presently thought to be dominated by homogeneous gas
phase reactions. In the case of atmospheric nitrates, a
particularly significant production pathway is through a reaction
between the hydroxyl free radical and nitrogen dioxide resulting
in nitric acid (HON02) formation. The fate of nitric acid in
the atmosphere is not well understood, though a portion of
gaseous nitric acid is known to enter into an equilibrium
with ammonia (NH3) to form particulate ammonium nitrate
(NH4N03). Present knowledge provides little support for
liquid phase oxidation as an important pathway to NOX
transformation.
2.3 Trends in Precipitation Composition and Deposition
Establishing trends in the chemical concentrations in
precipitation or the deposition of these materials in eastern
North America over the past 20 to 30 years is difficult
because the appropriate continuous data sets do not.exist.
Various stations or networks have been established by different
agencies for different purposes at different times during
this period. However, no one network has been operated for
more than a few years, and any trend must be established by
interpolation and extrapolation of the existing data sets.
-------
2-4
All the factors that can contribute to the difficultly in doing
this are pointed out in Chapter 2 of Report No. 2-14. Notwith-
standing these difficulties, the data do suggest an expansion
of the region covered by acidic rainfall, especially into
the southern and western U.S. There are differences of opinion
on whether the acidic deposition has actually increased in
amount, although on balance the data support the continuation
of elevated acidity levels at many locations in the northeastern
U.S. and eastern Canada.
Its becomes even more difficult to relate the suggested
trends to changes in emissions because the emissions were
not nearly so well documented in the 1950s and 1960s before
the statutory requirements of the clean air legislation
in the U.S. and Canada. The best estimates indicate a
moderate increase (approximately 20-40%) for SC>2, and a
dramatic increase (approximately 300%) for NOX emissions
between 1940 and the mid-1970s in the U.S. However, other
characteristics of the emissions have also changed; such as
the removal of a greater portion of the particulate loading
since 1971 and the steadily increasing emission heights in
recent years.
The Modeling Subgroup has focused its modeling efforts
on the present day situation using the best current estimates
of emissions and evaluation against the current multi-network
deposition measurements. These models are not being applied
against past data because of the uncertainties in past measure-
-------
2-5
merits and input data. Thus, trends are a non-issue as far as
the Work Group 2 modeling is concerned. However, it is
important to better establish the existence (or non-existence)
of past trends and to understand their significance and,
therefore, further work needs to be carried out. As for the
future, a strong commitment by the appropriate agencies in
Canada and the U.S. to continuation of monitoring networks
is required to ensure that we will not be faced with a lack
of appropriate data 10 and 20 years from now.
2.4 Seasonal Dependence of Atmospheric Deposition and Chemical
Transformation Rates for Sulfur and Nitrogen Compounds
A literature survey has been carried out into the seasonal
variations of the wet and dry deposition rates, as well as
the chemical transformation rates, of sulfur and nitrogen
oxides, with particular reference to deposition and transformation
parameters of relevance to long-range transport models. Both
relevant theoretical and experimental results have been
considered although a critical evaluation of the literature
has not been attempted.
From a theoretical view point, the deposition and trans-
formation rates of sulfur and nitrogen compounds could
potentially have a substantial seasonal variation. However,
it is difficult to draw conclusions about the magnitude of
this variation with any degree of confidence from the current
theories, with the possible exception of the wet and dry
-------
2^6
deposition of sulfur dioxide and the photochemical component
of its chemical transformation rate. Therefore/ the available
field data were also considered, although these were often
too scanty to be of much assistance.
An attempt was made to summarize the available information
on the seasonal variation of transformation/deposition rates
for the sulfur compounds. It was not intended to recommend
these values for use by the long-range transport modeler -
much more experimental and theoretical work is needed before
this will be possible - but rather, to indicate whether
seasonal changes in the parameter of interest are expected to
be greater or less than an order of magnitude. At present,
little more than this can be done. The conclusions are as follows;
1. The scanty available data suggest that the washout
rates of sulfates (and probably nitrates) should be
comparable in summer and winter. The rainout rates
could be strongly dependent on storm type, and hence
the time of year, because of the different mechanisms
whereby particles can be incorporated into precipita-
tion ( - some data suggest variations of an order-
of-magnitude or more).
2. Experimental results and theoretical considerations
suggest a seasonal variation of the wet scavenging
coefficient for sulfur dioxide which can be up to
several orders of magnitude, depending on the latitude.
-------
2-7
This variation is most pronounced in the northern
parts of America which receive appreciable amounts
of snow in the winter and convective storms in the
summer. Probably the same conclusions also apply
to nitrogen dioxide. Nitric acid vapor, being
highly reactive with all kinds of surfaces/ is
expected to show a smaller seasonal dependence of
the scavenging coefficient.
3. The situation is too confusing at present to draw
any conclusions about the seasonal dependence of
the dry deposition rate for sulfates (or nitrates).
In the winter/ deposition velocities would seem to
be 0.2 cms~l or less, but values reported for summertime
conditions range over an order of magnitude, including
negative numbers.
4. The dry deposition velocity of sulfur dioxide is
expected, from available experimental and theoretical
results, to show only a modest seasonal variation -
generally, less than a factor-of-two or so in any
given area. The same is probably true of nitrogen
dioxide and nitric acid vapor.
5. The gas-phase homogeneous component of sulfuric and
nitric acid formation rates is relatively well
understood, and has a strong seasonal variability,
especially at the northern latitudes. However, our
knowledge of the heterogeneous component, including
-------
2-8
in-cloud processes/ is too poor at present to allow
any conclusions regarding the seasonal dependence
of the overall chemical transformation rate of
sulfur and nitrogen oxides.
6. For many of the parameters under consideration,
during the winter months/ rates are strongly depen-
dent on latitude - e.g., photochemical conversion
rates of sulfur and nitrogen oxides above 45°N become
negligible, as do also wet deposition rates of gases
such as sulfur dioxide (because precipitation is largely
in the form of dry snow). This indicates that not
only the seasonal, but also the spatial variability
of deposition and transformation rates should be taken
into account in long-range transport models. Although
it may be too early to speculate, the following
approach does not seem unreasonable: during the summer
months, one might assume, as a first approximation,
the same values for deposition/transformation parameters
regardless of location, for each species of interest.
During the winter months, while rates at the southerly
latitudes might stay roughly the same as those in the
summer/ the models would include a dependence of
deposition/transformation on latitude/ which could
be quite pronounced for some of the parameters (such
as wet deposition of sulfur dioxide).
-------
2-9-
7. For the sulfur compounds, more experimental data
are badly needed, both under summer and winter-
time conditions, particularly on wet and dry
deposition rates of particulates and chemical trans-
formation rates in regional scale air masses (as
opposed to chimney plumes). Very little is also
known about in-cloud transformation and deposition
processes. For the nitrogen compounds, data are
required in almost every area of interest, and
immediate support for laboratory and field investi-
gations into deposition and transformation rates
of the major species (NO, N02/ HNC>3, nitrates and
PAN) is strongly recommended.
2.5 The Global Distribution of Acidic Precipitation and Its
Implications for Eastern North America
A review of the world-wide data on precipitation pH in
remote and exposed mid-latitude west-coast areas indicates
that all precipitation contains acidic materials in small
quantity. In the absence of any neutralizing alkaline
components, the acidic materials are sufficient to reduce the
pH to a value of about 5.0 and in some cases less. Nowhere,
though, are pH values in remote areas as low as those found in
the most acidic precipitation areas of northeastern U.S. and
western Europe. Several authors are now suggesting that the
reference level of 5.6 (the pH of rainwater in equilibrium
with atmospheric 62) is not appropriate and that departures
-------
2-1-0
from a value of near 5.0 would indicate the regional and local
modulations of the influence of "global background".
While pH is a useful single number that characterizes
the precipitation, it is the total deposition of acidity (H+
ions) and sulfate that is important in assessing the effects on
sensitive ecosystems. The deposition is the product of the
concentration and the rainfall amount. Thus in considering
the relevance of the low pH values in remote areas, this
must be considered. Sensitivity, in the form of the buffering
capacity of the receptor surfaces is also important in defining
the seriousness of impact. Most remote areas, especially
arid regions, are well buffered and so the impact of any
acidic deposition is minimized. In contrast, the regions
with lowest pH and highest depositions of H+ ion in the
northeastern U.S., eastern Canada and southern Scandinavia
cover large areas of poorly buffered lakes and soils and
thus have a major impact on the receptors there.
While there is considerable variability in the background
pH values, they are, in general, consistent with the concepts
proposed. The limited vertical profiles available also are
supportive of the hypothesis that most precipitation starts
off as acidic cloud droplets. It must also be pointed out
that some of the observations cannot be readily explained and
clearly, more analysis of existing data bases are required.
For example, trajectory analyses could be used to identify
-------
2-1-1
whether observed background levels are due to natural sources
or residuals from man-made sources far upwind.
The number of remote stations that have been established
9
in recent years is now beginning to generate substantial data
relevant to the issue of establishing and understanding the
background levels of air and precipitation chemistry. Rather,
than establishing many more such stations, the priority should
be to analyze and interpret the existing data base.
Some specific recommendations are as follows:
wherever possible at precipitation chemistry
stations, sampling should be done on an event
or at least on a weekly basis.
the precipitation chemistry data is much more
valuable and can be interpreted more readily
if concurrent basic air chemistry measurements
are made by filter-pack sampling.
more observations of the vertical distribution
of precipitation chemistry (and where possible
air chemistry) are needed. This can be done in
two ways
a) at mountain sites
b) with instrumented aircraft
-------
2-12
continuing efforts are required to refine the
estimates of natural emissions of acid compo-
nents into the atmosphere (they are presently
»
less accurate than estimates of man-made
emissions/ yet are equally important on the
global scale).
estimates (however approximate) are required
for emissions of the most important alkaline
materials into the atmosphere; at present
none exist.
-------
Chapter 3
MONITORING RESULTS AND INTERPRETATION
3*1 North American Precipitation Acidity
Precipitation chemistry was monitored at approximately ninety
locations in North America by the governments of the United States
and Canada in 1979. The observed annual mean precipitation-weighted
pH distribution is shown in Figure 3-1. The annual-mean pH of
precipitation was lowest in the eastern half of the continent.
The lowest pH (highest acidity) of about 4.2 occurred in a corridor
stretching through Ohio and Pennsylvania into southern Ontario.
The zone of maximum acidity is immediately downwind of large
sources in the upper Ohio River Valley. During periods of preci-
pitation, the prevailing flow is generally from the southwest in the
eastern U.S. and Canada. This situation facilitates the transport
of pollutants from major sources to the northeastern U.S. and
southeastern Canada.
A zone of high pH at mid-continent to the east of the Rocky
Mountains indicates the simultaneous absence of large acidic-
pollution sources in the region and the presence of alkaline
soil. Therefore wind-blown dust acts to cause more basic rain.
This situation is only typical of dry mid-continental areas where
abundant sources of alkaline materials are available.
On the west coast of the continent, precipitation is slightly
acidic (annual - value of about 5.6). The cause of that acidity
is not well understood. Possible explanations include: anthro-
pogenic sources which do exist, the release of biologically-
-------
.30"
^Figure 3-1.
Annual Average pH of Precipitation in North America
During 1979 Based On Observations By Canadian APN
and CANSAP Networks and American MAP3S and NADP Net-
works. (Note: An isoline is dashed where uncertainties
in its position are great due to lack of data.)
Issued by the Atmospheric Environment Service May 1981.
110
100'
90"
j
HO*
-------
3-3
produced organic sulfur compounds from the Pacific Ocean surface
that ultimately oxidize to sulfur dioxide and sulfuric acid in
the absence of neutralizing materials.
Further east/ the west coast mountains in North America
influence the continental pH distribution by acting as a cleansing
barrier. Strong precipitation scavenging induced by orographic
lift causes the air mass impinging from the west to produce
precipitation with a pH close to 5.6 in the absence of alkaline
wind-blown dust. This is borne out by observed pH in the vicinity
of the mountains of about 5.6.
To give an overview of the hemispheric acidity pattern, an
estimated pH distribution is shown in Figure 3-2 (Gravenhorst,
et al. 1980). Of note are the high precipitation acidity in
Eastern North America, Western Europe and Japan, and the more
alkaline precipitation in the large continental areas.
3.2 Temporal Variations
Air
In 1979, routine monitoring of the daily average concentration
of sulfur dioxide, sulfates and precipitation chemistry was
carried out at rural, regionally representaive sites (Figure
3-3) in eastern Canada (Barrie et al., 1980). The network is
the Canadian Air and Precipitation Monitoring Network (APN).
Monthly mean concentrations of sulfur dioxide and particulate
sulfate are given in Figure 3-4. On an annual basis, the lowest
-------
3-4
Figure 3-2. Estimated Distribution of the pH For Rain Water
in the Northern Hemisphere. Source: Gravenhorst,
et al.(1980).
-------
3-5
APN
KEJIMKUJIK
txPERIMENTA
LAKFS AREA
PRESENTLY OPERATING
O PLANNED
Figure 3-3.
Sites in the Canadian Air and Precipitation
Monitoring Network (APN) At Which Daily Air
and Precipitation Samples Are Collected.
-------
3-6
op
I
O
-
I
O
15 -
10 -
LONG POINT
CHALK RIVER
KEJIMKUJIK
ELA-KENORA
SULPHUR DIOXIDE
N
Figure 3-4,
Temporal Variations of the Monthly Average Con-
centration of Atmospheric Sulfate and Sulfur Di-
oxide At APN Sites During 1979.
-------
3.-7
concentrations occurred at Experimental Lakes Area (ELA)-Kenora
-J -3
in northwestern Ontario (S02 = 1.5 jug/m ; S04 = 1.6 jjg m~°).
In the mid-latitudinal westerlies, ELA is upwind of major North
American sulphur sources. Long Point, on the other hand is
located in an industrialized section of Ontario and is northeast
of major emissions in Ohio and Pennsylvania. At all sites in
eastern Canada, there are seasonal variations of atmospheric
sulfur dioxide levels. The levels are highest in the winter and
lowest in the summer. On a percentage basis the amount of variation
about the annual mean concentration depends on the location. It
is lowest in source regions and highest at samplers which are remote
from source regions. The percent standard deviation about the
annual average concentration is 43 af Long Point on Lake Erie,
63 at Chalk River (500 km away), 75 at Kejimkujk and 108 at ELA,
the furthest location from the major source area of emission.
Particulate sulfate concentrations are highest in summer at
all APN sites except ELA-Kenora. A winter maximum in both sulfur
species at Kenora owes its existence largely to meteorological
factors. Maximum transport westward from eastern North America
in winter coincides with a large winter peak of background sulfur
concentrations in artic air masses (Barrie et al., 1981) that
prevail at this mid-continental location during winter.
At APN sites located in the continental pollution plume, the
sulfate seasonal cycle is one hundred eighty degrees out-of-
phase with the sulfur dioxide cycle. Indications are that the
-------
.3-8
summer sulfate maximum owes it existence to a summer maximum in
the conversion rate of sulfur dioxide to sulfate. One
manifestation of higher S02 to sulfate conversion in summer is
that, at all stations/ the ratio of S02 to total airborne sulfur
(S02 + 304) tends to be lowest in summer (e.g. see Figure
3-5a for Long Point).
On a daily basis, the concentrations of sulfur oxides are
highly episodic regardless of location in eastern Canada. Polluted
and non-polluted periods of 3 - 6 day duration alternate regularly.
The dry deposition rates of these acidic substances, which is
thought to be roughly proportional to their atmospheric concentra-
tions, is equally episodic. A comparison of monthly wet and dry
deposition rates of sulfur compounds is made later in this chapter.
Recent work in the U.S. concerning transport of particulate
sulfate has been reported by Paresh and Husain (1981). In this
study, particulate sulfate in air was monitored continuously from
June 1978 through December 1979 at Whiteface Mountain, New York.
=
The influx of transported 804 was evaluated by sectors to assess
the relative contribution from the U.S. and from Canada. The
daily 304 concentrations were related to surface-air trajectory
ensembles. During the study period the site was influenced
approximately equally by the Canadian continental polar winds and
the U.S. maritime tropical winds. However, the maritime air masses
from the U.S. were the principal conveyors of very high urban-
like 304 concentrations at this site and transported about 4 to 5
-------
3-9
100 -
Psl
a
CE
cr
Lu
Ld
_J
a
UJ
_j
a
!! IT
cn
LTl
4-
PJ
a
t_n
0
2.00 -
.00
0.00
N D
Figure 3-5.
j'r'H'n'M'J1 .j'n'51 D'N'D
Top: The Fraction of Total Airborne Sulfur Exist-
ing As Sulfur Dioxide At Long Point On the North
Shore of Lake Erie on a Daily Basis (1978-1979).
Bottom: The Concentration of Total Airborne Sulfur
At Long Point On a Daily Basis. (Note the preval-
ence of episodes of elevated sulfur levels of 3-6
davs duration.)
-------
3-10
times more 804 than did the continental polar air masses from
Canada.
Precipitation
The temporal variations of monthly precipitation-weighted
mean concentrations of the four major ions (H+, NH4+, S04 and N03~)
at APN sites during 1979 are shown in Figure 3-6.
At sites nearest to the major pollutant source region (i.e.,
Long Point and Chalk River), sulfate concentrations were highest in
the summer half of the year and lowest in the winter half, while
hydrogen and nitrate ion concentrations showed no significant
seasonal variations. A similar seasonal variation of sulfate
and nitrate in precipitation has been observed south of the border
in the American northeast in the MAP3S network(MAP3S/RAINE,1981)
(see Figure 3-7). However, the seasonal variation of hydrogen ion
follows the sulfate in the MAP3S data. At ELA-Kenora, a site more
remote from the influence of continental sources, sulfate did not
peak in any particular season, nitrate concentrations varied little,
but precipitation acidity was highest in winter. The latter was
accompanied by a minimum in ammonium ion concentration and also
coincides with the time of year when wind-blown dust would be a
minimum. On an equivalent basis, sulfate was more abundant than
nitrate throughout the year at all stations. An exception was
at Chalk River when they were equal in concentration during the
winter.
-------
400
200
LONG POINT
3-11
H+
------ NO;
T 1
200
^
cr
3 100
CO
m 0
Z
o
S BO
z
o
CHALK RIVER
i I
LU
>
X
Z
O
40
0
KEJIMKUJIK
Figure 3-6,
N
The Temporal Variation of Monthly Precipitation-
Amount Weighted Mean Concentrations of Four Major
Ions At APN Sites During 1979 (see Fig. 3-3).
-------
3-12.
/u+\
8-1
O
o
c
o
O
X
JASON DUFMAMJJASON DU F M A M J
1977 1978 1979
s-
(S04=)
LEGEND
= Wh!t«face
Ithaca
. + = P«nn Stat«
x = Vlrglnlo
A. Os III! no I s
.'+;. .O _ * V = 8rookhav«n
S. .x « .'.'«'*. ' B = D«lawar«
______ __x = 0hlo
O
O
o
JASONDUFMAMJJASONDIJFMAMJ
1977 1978 1979
O-,
oo
o
C o
(N03 )
v A" ' ' 7 ^ '-
X. -f X .»-._ * n ' * A-
O
«-*
O
O
"0
1 I r T i T I r
JASONDIJFMAM J JASONDlJFMAM J
1977 1978 1979
FIGURE 3-7 Monthly Mean Deposition-Weighted Concentrations for
H+, S04 and ~
-------
3-13
Wet Versus Dry Deposition
Monthly wet and dry deposition of sulfur oxides were
derived from APN air and precipitation data (see Figure 3-8).
Dry depositions was calculated from air concentrations measured
by the APN network (Figure 3-2) and from the deposition velocities
listed in Tables 3-1 and 3-2. Deposition velocities were estimated
using monthly frequency distributions of Pasquill-Gifford stability
classes and the techniques of Sheih et al. (1979), modified to
include higher particulate sulfate surface-resistances. These
values can be considered only rough estimates of the dry deposition,
Wet deposition was calculated from the product of precipitation
amount and ionic concentrations (Figure 3-6).
In general, dry deposition of sulfur is not negligible
compared to wet. The ratio of dry to wet deposition is highest
for locations that are close to source regions. It is also
higher in winter than in summer since in winter, SC>2 concentrations
near the ground are highest and precipitation amount is usually
lowest.
3.3 Classification According to Air Parcel Origin
Measurements of daily mean concentrations of particulate
major ions and sulfur dioxide made at APN sites (Figure 3-2) in
1979 were classified according to air parcel origin.
Back-trajectories at a level of 925 mb were calculated for
every site and for four times (0, 6, 12, 18 'GMT) each day for a
period of five days based on six hour time steps within the
-------
LONG POINT SO,
3-14
\
j i
I J I I I I
C\J
E
o
O
O
Q.
LU
Q
cc
D
Q.
CO
z
o
4r-
CHALK RIVER
X
X
x
J I
I I
KEJIMKUJIK
r
1 -
Figure 3-8,
N
D
The Temporal Variation of Monthy Wet and Dry Dep-
osition of Oxides of Sulfur At APN Sites (see Fig.
3-3). Dry Deposition is Calculated From Air Concen-
trations (Fig. 3-4) and Deposition Velocities (Tables
3-1 and 3-2). Wet Deposition is Measured.
-------
3-15
Table 3-1. Estimated Monthly Average Dry Deposition
Velocities (cm s~l) of Sulfur Dioxide
at AEN Sites
SITE JFMAMJJASOND
ELA 0.15 0.15 0.45 0.43 0.43 0.43 0.42 0.42 0.41 0.43 0.43 0.15
Long Point 0.15 0.15 0.71 0.68 0.64 0.41 0.37 0.37 0.55 0.58 0.65 0.15
Chalk River 0.18 0.18 0.29 0.27 0.22 0.22 0.18 0.27 0.22 0.22 0.23 0.18
Kejimkujik 0.18 0.18 0.44 0.37 0.37 0.36 0.30 0.30 0.25 0.28 0.29 0.17
Table 3-2. Estimated Monthly Average Dry Deposition
Velocities (cm s"1) of Particulate Sulfate
at APN sites.
MCNTH
SITE JFMAMJJASOND
ELA 0.08 0.08 0.19 0.19 0.20 0.15 0.15 0.15 0.14 0.15 0.14 0.08
Long Point 0.08 0.08 0.10 0.10 0.09 0.08 0.08 0.08 0.10 0.10 0.11 0.08
Chalk River 0.10 0.10 0.32 0.29 0.27 0.30 0.30 0.29 0.18 0.18 0.19 0.10
Kejimkujik 0.09 0.09 0.33 0.33 0.34 0.24 0.25 0.23 0.17 0.17 0.17 0.09
-------
3-16
5-day period. However, only the last two days of the trajectories
were used in the classifications since most major sources are
within two days travel time of the locations under consideration
and the accuracy of trajectories decreases after two days.
For this preliminary sector analysis, five concentration
ranges were established for each chemical species in air and
precipitation. The compass was divided into eight equal sectors.
Four 2-day back-trajectories were used to determine the sector froi
which most of the air originated in a 24-hour period. Each
observed daily concentration was classified into a particular
concentration range (Table 3-3) and compass sector.
The trajectory-wind-roses at APN sites were plotted on
a map of North America (Figures 3-9 to 3-13). There is a differenl
map for each chemical species. In addition to the frequency
of occurrence of air from any sector being represented by the
length of the radial projection for each width, the wider the
projection the higher the concentration of each chemical species.
Concentration ranges are listed in Table 3-3 for particulate
sulfate, sulfur dioxide, pH, precipitation sulfate and precipita-
tion nitrate.
The reader is cautioned that the reported trajectory-wind-
roses for precipitation days (Figures 3-11 to 3-13) may contain
uncertainties. Precipitation events are often accompanied by
frontal passages, the presence of which can reduce the accuracy
of an air parcel trajectory analysis. Nevertheless, it is felt
-------
3-17
Table 3-3 Definition of Concentration and pH Ranges
' for the Trajectory Cases
Particulate Sulfate : bdl* -
2 -
4 -
10 -
>
Gaseous Sulfur Dioxide : bdl
3 -
6 -
15 -
>
Rain pH : 3.80
(Note that in Figure 3-11 the 3.80 -
lower pH ranges correspond 4.40 -
to wider radial projections) 5.00 -
Rain Sulfate : bdl -
2 -
4 -
8 -
>
Rain Nitrate : bdl
1.29 -
2.58 -
5.17 -
*below detection limit >
2 jug/m3
4
10
20
20
3 /ag/m3
6
15
30
30
4.40
5.00
5.60
5.60
2 mg/1
4
8
16
16
1.29 mg/1
2.58
5.17
10.33
10.33
-------
Figure 3-9,
Trajectory Wind Roses At the APN Sites For Part-
iculate Sulfate Based On Two Day Back Trajectories
At 925 mb(about 800 meters) During January - Dec-
ember 1979. (Scales: 9mm = 10% and see Table 3-3)
oo
M> o $;, o^u \
-------
Figure 3-10. Trajectory Wind Roses AtlH^N Sites For Sulfur Dioxide
Based On Two Day Back Trajectories At 925 mb(about
800 meters) During January - December 1979. (Scales:
9mm = 10% and see Table 3-3)
Ul
I
-------
Figure 3-11.
Trajectory Wind Roses At^^N Sites For Rain pH Based
On Two Day Back Trajectories At 925 mb(about 800
meters) During January - December 1979. (Scales: 9mm
= 10% and see Table 3-3)
" ' :tr^
-------
Figure 3-12. Trajectory Wind Roses At^B>N Sites For Rain Sulfate
Based On Two Day Back Trajectories At 925 mb(about
800 meters) During January - December 1979. (Scales:
9mm = 10% and see Table 3-3)
*- ( :»"3
-------
Figure 3-13.
Trajectory Wind Roses At^PPN Sites For Rain Nitrate
Based On Two Day Back Trajectories At 925mb(about
800 meters) During January - December 1979. (Scales:
9mm = 10% and see Table 3-3)
U)
I
ro
NJ
r^>^>r//'o r(* «
\'~t JL'H/- -I'-o'
-------
3-23
that the graphs in Figures 3-11 to 3-13 are accurate enough to
allow one to draw some valuable conclusions.
The trajectory-wind-rose data contain a great deal of
information pertinent to the long-range transport of pollutants.
Some of the most important results are summarized as follows:
(1) in general/ the highest concentrations of acid-related
substances in air or precipitation occur when air originates
from a sector between southeast and northwest; the lowest for
air from a sector between north and southeast;
(2) southerly and southwesterly trajectories are more prevalent
during precipitation periods than during the whole year
(compare the envelopes of trajectories in Figures 3-7 and 3-9);
(3) the phenomenon of long-range transport is clearly demonstrated
by results at Kejimkujik National Park in Nova Scotia which
is downind of and remote from continental sources. The highest
concentrations in air and precipitation occur mostly when air
originated from the southwest to west sector.
In the U.S., the Air Resources Laboratories' Atmospheric
Transport and Dispersion Model (ARL-ATAD) has been used to develop
a trajectory climatology in conjunction with the precipitation
chemistry at MAP3S Whiteface Mountain and Illinois sites (Wilson,
et al., 1980). Trajectories were calculated for all of 1978 and
1979 and weighted by precipitation amounts during every six-hour
period to determine a dominant direction or origin of air mass
for each precipitation event. Each event was then classified in
one of twelve sectors covering 30° each. Only the final two
days (48 hours) of approach trajectory end points were considered.
This allowed differentiation between so-called "Ohio Valley/Midwest"
and "Canadian/Great Lakes" air masses. The so-classified wet
ion deposition as a function of trajectory sector are shown in
Figures 3-14 through 3-16. The Ohio Valley/Midwest events were
-------
3-24
OHIO VALLEY/MIDWEST-
CANADIAN/GREAT LAKES-
TOTAL PRECIPITATION-
17.3 LITER
7.9 LITER
30.6 LITER
26X OF TOTAL.
ee
T
ee
1 2Q
i se i eo 21 a 243 270
TRAJECTORY SECTOR
see
b)
te -
OHIO VALLEY/MIDWEST-
CANADIAN/GREAT LAKES-
TOTAL DEPOSITION-
23.3 MG/M2
11.7 MG/M2
37 . 4 MG/M2
3IX OP TOTAL.
OHIO VAL.L.EY/
MXOWEST
CANADIAN/
GREAT UAKCS
tee i eo 2 IB 2-*e 270
TRAOECTORY SECTOR
see 338 see
Figure 3-14.
Top: Precipitation Volume, and Bottom: Hydrogen
Ion Total Wet Deposition, At Whiteface Mountain,
New York, Based On Two Day Back Trajectories In
the Boundary Layer Assigned To 30° Sectors.
Source: Wilson, et al, 1980.
-------
eeei
see-
A
N
£
\
O
£
Z see
o
H
h
H
8
0. aecr
UJ
O
o
w i ee-
u
3-25
OHIO VALLEY/MIDWEST-
CANADIAN/GREAT LAKES-
TOTAL DEPOSITION-
1087.9 MG/M2
531.6 MG/M2
1700.5 MG/M2
64 X OF TOTALv
3IX OF TOTAL.
I
60
I
98
I 1 T
120 150 180 210 248 278
TRAJECTORY SECTOR
308 338 360
OHIO VALLEY/MIDUEST-
CANADIA'N/GREAT LAKES-
TOTAL DEPOSITJON-
66X OF TOTAL.
646.7 MG/M2
283.4 MG/M2
994.6 MG/M2
28 X OF TOTAL.
Figure 3-15.
CANADIAN/
GREAT LAKES"
. x x x y
Tr
\ 6O ISO 210 2-40 270
TRAJECTORY SECTOR
300 330 aea
Top: Sulfate, and Bottom: Nitrate Total Wet Dep-
osition, At Whiteface Mountain, New York, Based
On Two Day Back Trajectories In the Boundary Layer
Assigned To 30° Sectors. Source: Wilson, et al, 1980,
-------
ee ~\
4B
X 30
O
I
s/
0
M
v>
O
Q.
Ul
Q
O
u
3-26
OHIO VALLEY/MIDUEST-
CANADIAN/GREAT LAKES-
TOTAL DEPOSITION-
Oe.3 MO/M2
17.7 MOXM2
83.7 MO/M2
72X OF TOTAL
OF TOTAL
~f
60
122
tea tea 210
TRAJECTORY
T T
24Q 27Q
SECTOR
sea 330 see
OHIO VALLEY/MIDWEST-
CANADIAN/GREAT LAKES-
TOTAL DEPOSITXON-
66X OF TOTAL
1 ee . e
ee.3
1 8-4 . «*
CANADIAN/
GREAT LAKES
MG/M2
37* OF TOTAL
t se i ee 21 o
TRAJECTORY
330 360
Figure 3-16.
Top: Calcium, and Bottom: Ammonium Total Wet Dep-
osition, At Whiteface Mountain, New York, Based On
Two Day Back Trajectories In the Boundary Layer
Assigned To 30° Sectors. Source: Wilson, et al, 1980
-------
3-27
defined as those having their origin in the sector 165° through
285°, while the Canadian/Great Lakes events were designated by
the sector 285°-030°. The wet deposition directional patterns
for all ions were found to closely follow the precipitation
variability in every case. The wet deposition cumulative total
of the Ohio Valley/Midwest sector was found to account for approx-
imately 50-70% of the total wet deposition that could be classified
into the 48-hour approach trajectories. Wet deposition from the
Canadian/Great Lakes sector resulted largely from winter coastal
storms in the form of snow or mixed precipitation. Even though
ionic concentrations were generally lower for these events/ the
large values or precipitation sample volume resulted in almost
30% of the t'otal wet deposition at Whiteface Mountain.
The same analysis procedure (ARL-ATAD model) was applied to
the MAP3S precipitation chemistry for the Illinois site during
1978. Since large pollutant sources for S02, and NOX lie to
the east or southeast of the Illinois stations, one might expect
to see a significant difference in the deposition pattern pollution-
related ions observed at Whiteface Mountain during the same year.
The cumulative wet deposition total at Illinois for these events
is plotted in Figures 3-17 and 3-18 revealing a very interesting
picture. Approximately 70% of the total precipitation for Illinois
was found to be associated with the southwest approach sector.
This is consistent with synoptic considerations whereby southwesterly
-------
o -
7 ~
K
LJ
s -
z
0
H
t-
<
t-
M
a.
H
0
LI
a
a.
s -
3-28
SOUTHWEST SECTOR-
TOTAL. PRECIPITATION-
IS. SI LITI-TW
25.31 LITFIR
1 20
1 T 1 r
150 163 210 24O 270
TRAJECTORY SECTOR
300 33Q 3(
4B -1
35
SOUTHWEST SECTOR-
TOTAL. DEPOSITION-
35 . I 3 MG/T-.2
44.85 MG/M2
78X OF TOTAL
O
I
0
H
O
0.
u
0
n
I
10
oa
GO I «W 2IO
Tl'AOt-'CTOKY
3eo
OR
Figure 3-17,
Top: Precipitation Volume, and Bottom: Hydrogen
Ion Total Wet Deposition, At Whiteface Mountain,
New York, Based On Two Pay Back Trajectories In
the Boundary Layer Assigned To 30° Sectors. Source
Wilson, et al, 1980.
-------
3-29
soon
4OCT
f\
N
I
o seer
I
0
H
H
o
0.
u
0
8 i**i
2
SOUTHUEST SECTOR-
TOTAL DEPOSITION-
64X Of TOTAL
574.22 MG/M2
903.04 MG/M2
30
I 2 Q
1 60 180 210 240 270
TRAJECTORY SECTOR
30O 330 363
SOUTHWEST SECTOR-
TOTAL DEPOSITION-
I 1 OA . 4 I MGXM'O
.11 MG/M2
OF TOTAL
Figure 3-18
Top: Sulfate, and Bottom: Nitrate Total Wet Dep-
osition/ At Whiteface Mountain, New York, Based
On Two Day Back Trajectories In the Boundary Layer
Assigned To 30° Sectors. Source: Wilson, et al, 1980,
-------
3-30
flow normally precedes an approaching cyclone system and subsequent
deposition events. The reminder of the precipitation was found
to be essentially equally distributed over the various sectors
with the exception of the 0-30° segment.
Air masses approaching the Illinois site from the southwest
were found to deliver about 70% of the total annual precipitation
resulting in the deposition of approximately 70% of the major
ions (Figures 3-17 and 3-18). The analysis of precipitation
chemistry by air mass at both sites showed:
Whiteface Mountain (1978)
Midwest/Ohio Valley 56% of the annual precipitation delivering;
(160°-280° sector)
62% of the annual [H+] deposition
64% of the annual [864] deposition, and
65% of the annual [NC>3~] deposition
Canadian/Great Lakes 26% of the annual precipitation delivering:
(280°-030° sector)
31% of the annual [H"1"] deposition
31% of the annual [804] deposition, and
28% of the annual [N03~] deposition
Illinois (1978)
(160°-280° sector) 71% of the annual precipitation delivering:
78% of the annual [H+] deposition
67% of the annual [804] deposition, and
64% of the annual [NO-j"] deposition
Wilson, et al. (1980) concluded the following:
-------
3.-31
(1) precipitation volume, more than any other single factor/
determines the amount of deposition for the three pollution-related
ions, [H+] , [S04=] and [N03~]; (2) deposition of ions from
the Midwest/Ohio Valley air and Great Lakes/Canadian air does
not reflect the very significant differences in emissions that
are located in these two regions; (3) the Illinois results further
substantiate the above conclusion; (4) there seems to exist no
simple, straightforward relationship between emission source(s)
for acid precipitation precursor gas(es) and receptors of "acid
rain," i.e., [H+], [S04=] and [N03~] ions; (5) the chemical
transformation pathway(s) seem to be complex and insensitive to a
48-hour back trajectory analysis presented here and based upon
available meteorological and chemical information; and (6) the
concept of a "superbowl" (synoptic scale air shed) would explain
to some extent the uniformity in directional variability of
annual wet ion deposition in that the final products, i.e.,
acidic material is being rather evently deposited over a very
large region.
These conclusions should be checked at Canadian and other
eastern U.S. monitoring sites. A summary of the major source-
receptor analyses using trajectories or storm tracks is provided
in Table 3-4.
-------
3-32
Table 3-4. Major Source - Receptor Analyses Using Trajectories or Storm Tracks
PRINCIPAL AUTHOR
POLLUTANT
LOCATION
CONCLUSION
SAMSON (1980)
S04
WHITEFACE STRONGLY CORRELATED WITH SOUTH-
MOUNTAIN WEST WINDS AND STAGNATION _> 36
HOURS BEFORE UPWIND
WILSON, ET AL,.
(1980)
H+ AND
WET SO,
WHITEFACE STRONGLY CONTROLLED BY PRECIPI-
MOUNTAIN TATION AMOUNT AND CORRELATED
WITH SOUTHWEST WINDS
PAREKH AND HUSAIN 804
(1981)
WHITEFACE MARITIME TROPICAL WINDS (FROM
MOUNTAIN U.S.) CONTRIBUTE 4-5 TIMES MORE
THAN CONTINENTAL POLAR WINDS
(FROM CANADA)
BARRIE, ET AL.,
(1981)
AND WET S04
AND WET N03
6 SITES GENERALLY HIGHEST VALUES FOR
IN EASTERN TRAJECTORIES FROM THE SOUTH-
CANADA EAST TO NORTHWEST SECTOR
MERRITT (1976)
40 TRACE
ELEMENTS
CHALK RIVER
(CANADA)
GENERALLY HIGHEST CONCENTRA-
TION FOR STORMS TRACKS FROM
THE SOUTHWEST
-------
3-33
3.4 Episodes of Lorig-Rarige transport
Episodes of elevated pollutant concentrations over a regional
area have been documented in several studies (c.f. Whelpdale, 1978;
Chung, 1977; Mueller et al.,*1979; and Lyons et al./ 1980). One
of the most common meteorological situations responsible for
regional episodes is the large slow-moving high pressure system
or anticyclone. Over eastern North America, these occur in summer
or winter. They usually contain extremely hot air in the summer
and very cold air in the winter. Consequently, pollutant emissions
related to either air conditioning or space heating are high.
Typically, a low level inversion develops that traps pollutants
in the lowest 1-2 km of the atmosphere, thereby preventing their
dispersal. In summer, the intensity of solar radiation is
sufficient to cause considerable photochemical activity in the
relatively cloud-free high pressure area and leads to elevated
ozone concentrations. A particular episode is described in
detail below.
In February 1979, a winter regional pollution episode
occurred in eastern North America in conjunction with a cold
high pressure system. It produced extremely high concentrations
of sulfur dioxide and particulate sulfates in eastern Canada.
Regional pollutant levels in the air mass entering southwestern
Ontario were so high that industries were requested by the Ontario
government to reduce emissions so as to not exacerbate the problem.
The episode was monitored in eastern Canada at rural sites in the
-------
3-34
APN Network (Figure 3-2) by Environment Canada and in southern
Ontario by the Ontario Ministry of Environment. Reports of the
incident were published independently by Barrie et al. (1980) and
Shenfeld et: a 1. (1980).
The cold air mass comprising the high pressure zone originated
in northwestern Canada. The center of the high pressure zone
moved from mid-continent progressively eastward until it was in
the position shown in Figure 3-19 and 20 on February 19 and 20,
respectively. The air mass was very stable with an intense
noctural ground-based inversion and overruning warmer air aloft.
In eastern Canada elevated concentrations of oxides of
sulfur (Figure 3-21) occurred first at ELA-Kenora in northwestern
Ontario on February 17 as the centre of the high pressure zone
passed east of the site. As the high pressure zone approached,
daily mean pollutant concentrations increased from relatively
low values (0.76 jug m S04= and 1.1 jug m~3 S02) to maximum
values on February 18-19 (4-9 pg m~3 S04= and 18-29 /ag m~3
S02) as the southerly flow to the west of the center of the
high (see Figure 3-19) carried pollutants from the lower Great
Lakes region a distance of over 1000 km to the site. Air parcel
back-trajectories are shown in Figure 3-22.
In southern Ontario, the pollution episode peaked on February
20, one day later than in northwestern Ontario. At this time,
the high pressure zone was positioned on the east coast (Figure
3-20) giving rise to a southerly flow over eastern North America.
-------
Figure 3-19. Surface Weather Map Sho^Wg Large High Pressure
Area Centered Over the Midwestern States On
February 19, 1979.
1000 x IQI2 1016 ,
u>
I
GO
Ul
io?:o
GV.
-------
Figure 3-20.
1012
1015
1020
Surface Weather Map Showing the Movement of the
Large High Pressure Area From the Midwest (see
Figure 3-19) To the East Coast On February 20, 1979,
1020 ,1016 ^ 1016
1C 7.0
IOIG
u>
I
U)
en
-------
3-37
NDV DEC UHN FEB MRR flPR MRY dUN
Figure 3-21.
Top: Daily Sulfate,and Bottom: Sulfur Dioxide
Concentrations (jug m~3) At the Experimental Lakes
Site (see Figure 3-3).
-------
3-38
'.<-,
LONG POINT I
Figure 3-22. Back Trajectories At 1000mb During February
16-20, 1979. (Crosses mark the 24-hour time
intervals)
-------
3-39
At Chalk River, Ontario (Figure 3-22), the daily mean concentrations
of sulfate and sulfur dioxide were 13.8 and 100 jug m~"3, respectively,
This sulfur dioxide concentration was the highest value observed
t
between November 1978 and June 1980. Air parcel back trajectories
are shown in Figure 3-22 for Chalk River and Long Point and indicate
an air parcel origin south of the Great Lakes. In southwestern
Ontario, daily mean S02 concentrations were 110 to 230 jug m~ .
Total suspended particulate (TSP) concentrations averaged 90 jug
m~"3, ranging from 30 to 130 ^g m~3. About 30 percent of the
TSP consisted of sulfates. Visibilities at southern Ontario
airports decreased from 20 km on the afternoon of the 19th (prior
to the incident) to 5-10 km on the 20th.
A light mixed rain and snow event (accumulation of 2 mm)
collected at Long Point on Lake Erie on February 20 during the
episode had a pH of 3.8. The concentrations of sulfate and
nitrate were 229 and 98 jug 1""^, respectively. In contrast,
precipitation collected on February 23 and 25 had a pH of 4.1
and 5.6, respectively.
Episodes of low pH values over a regional area have been
documented recently in the MAP3S/RAINE Oxidation and Scavenging
Characteristics of April Rains (OSCAR) Program. The objectives
design, and synoptic overview of the OSCAR program has been
described recently by Hales (1981). Results of an interesting
case study from the OSCAR program, when both the high and
intermediate density networks were activated, during the
-------
3-40
passage of a cold front with north-sourth orientation are shown
in Figure 23 and 24. The pH of the first rain sample at each
station during April 22-24, 1981/ shows a fairly organized
pattern with the minimum pH values over the same general area
as for the annual mean (see Figure 3-1). The change of pH
with time at four selected stations during April 22-24, 1981,
shows the pH values generally increased during the event at
all four stations. The large changes in pH values within a
short period of time is an especially interesting feature of
these data to explain from more detailed analyses of the
chemistry and meteorology.
3. 5 Background SuIfur _in_ North Ame^rica_ East of the Rocky Mqun t a ins
The models discussed in this report attempt to estimate S02
and sulfate concentrations and depositions from the spatial
distributions of emissions and wind and precipitiation fields.
For any given region area there are three basic sources of
atmospheric sulfur:
(1) natural sulfur (biogenic or wind-blown dust) released within
the modeling area;
(2) anthropogenic sulfur released within the modeling area';
(3) sulfur entering the modeling area from outside. This sulfur
may be of natural or anthropogenic origTn"! If" all anthropo-
genic sources within the modeling area were absent, the
concentration of oxides of sulfur in air and precipitation
would be by definition 'background' concentrations. Background
concentrations are caused by natural sources within and natural
and anthropogenic sources outside the region.
The modeling area of concern in this report is that portion of
North America east of the Rocky Mountains. Within this area
-------
Figure 3-23,
pH of the First Rain Sample At Each Station In
the OSCAR Program Network During the Storm Event
of April 22-24, 1981.
-------
Figure 3-24
i
Q.
Change of pH With Time A^Four Selected Stations
In the OSCAR Program Network During the Storm Event
of April 22-24, 1981.
Ji
1
5
4
1
5
4
2
5
4
C
^~
4
_
^
2
)4
/
1
!
5
1
3
[j '
>3
1
05
1 1 1 1 1 1 1 1 1 1 1 1 II
, i i , MIIRRARH RRHHk MH
_j ' APRIL 23-24
1 1 <
1 1 1 1 1 1 1 1 1 1 1 1 1 1 7
16 \7 18 19 20 21 22 23 24 01 02 03 04 05 06
UPTON, NY
ri APRIL 23-24
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
14 15 16 \7 18 19 20 21 22 23 24 01 02 03 04
SCOTTDALE, RA _
APRIL 22-23
i r ' ' ' ' f.
1 1 1 1 1 1 1 1 1 1 1 1 1 1 <
24 01 02 03 04 05 06 07 08 09 10 II 12 13 14
Ar-H . r-^~U
p 1 CHAMPAIGN, ILL
APRIL 22
1 1 1 1 1 1 1 1 1 1 1 1 1 III
06 07 08 09 10 II 12 13 14 5 16 \7 18 19 20
HOUR
10
I
-------
3.-43
anthropogenic sulfur emissions total 19.7 4- 6 Tg yr.""-1- while
natural emissions are 0.4 - 0.5 ± 0.1 Tg yr."1. A breakdown of
these emissions for eastern and western Canada and United States
is given in Table 3-5. The ratio of anthropogenic to natural
emission is about 24:1 in the east and about 16:1 in the west.
Therefore, within the modeling area, natural sources of sulfur
are unimportant compared to anthropogenic sources.
Air entering the modeling area also contains some sulfur.
In any model, the concentration of that sulfur can be an important
boundary condition, assuming that the "hemispheric background level"
is significant.
For eastern North America there are three major air-mass
sources; the Pacific Ocean, the Carribean and the Atlantic south
of 30° N and the Arctic. Background sulfur concentrations in
airstreams entering the modeling area from each of these sources
will be discussed in turn.
Pacific Air
Pacific air crossing the west coast will have concentrations
of close to "hemispheric background levels." As the air crosses
the coastal range, the orographic precipitation induced in an
air mass by the mountains acts as a scavenging curtain that
for the most part keeps sulfur originating in the Pacific basin
from reaching the lee of the Rockies. Thus, background sulfur
concentrations just to the east of the Rockies are extremely low.
Measurements in northern Alberta during summer (Barrie and.Barrie
-------
3-44
Table 3-5 A Breakdown of Natural and Antropogenic
Sulfur Emi_ss_i_qns (T?g LS .jyr._~1 )
East
West
Totals
CANADA
Natural
0.25!
0.25!
0.5
Anthropogenic
1.8
1.2
3.0
U.S.A.
Natural
0.40*
0.1 +
0.5
Anthropogenic
14**
2.7
16.7
! EPS (1980)
* Galloway and Whelpdale (1980)
** EPRI/SURE II
+ Estimated assuming that natural areal emissions between the
Rocky Mountains and latitude 95°W amount to no more than
twice the terrestrial emissions in the eastern region.
-------
3-45
et al. (1981) yield background sulphur concentrations of about
4 ug m~3 . Further south downwind of California on the lee of
the Rockies, concentrations are about 2 jug m~^ . It was pointed
out in Section 3.1 that one manifestation of the precipitation
scavenging over the coastal mountains is a pH value of 5.6 in
precipitation sampled downwind of the mountain. At west coastal
locations, onshore trajectories are associated with precipita-
tion pH's of about 5.0 which is probably reasonably representa-
tive of hemispheric background values.
Air from the Caribbean and Southern Atlantic
In contrast to air masses from the Pacific, nothing prevents
the pollutants in Caribbean and southern Atlantic air-masses
from entering the modeling area. In summer a common route for
this maritime tropical air-mass is up the Mississippi Valley and
then northeastward through southern Canada. Background sulfur
concentrations in this air remain a controversial subject.
There is some evidence suggesting that the concentrations of
"background" sulfates and sulfur dioxide are significant
(Reisinger and Crawford, 1981). Others contend that although
background concentrations are somewhat elevated relative to
the modified Pacific air masses they are insignificant compared
to the large anthropogenic emissions along the route commonly
travelled by this air-mass (Henry and Hidy, 1980).
-------
3-46
Arctic Air
Arctic air penetrates the modeling area along its northern
boundary most often during the period from November to April. In
mid-winter continental arctic air can penetrate as far south as
Florida on rare occasions but, more commonly, as far south as the
central eastern states.
The chemical content of continental arctic air has recently
been under intense study by many northern hemisphere (c.f.
Atmospheric Environment Arctic Issue, 1981) in connection with
a deterioration of visibility referred to as "arctic haze", which
is most pronounced in the winter season.
It has been found that anthropogenic particulates consisting
mainly of sulfates, soot and organic carbon emitted at mid-
latitudes are the cause of the haze. The atmospheric concentration
of these particulates as indicated by sulfate was an annual
cycle reaching a minimum value (close to zero) between June and
September and a maximum value of_2 ^ug/m^ between January and
March. This is illustrated in Figure 3-23 from measurments made
at in the Canadian arctic by Barrie et' a_l. (1981). In early
1980, weekly average sulfate concentrations ranged between 1.2
and 3. 5 jug m~^.
Particulate pollution in the arctic air mass is distributed
over large areas rather homogeneously. Little difference is
observed between sulfate concentrations at two sites in the
Canadian Arctic: Mould Bay and Igloolik, about 600 km apart
-------
3-47
(Figure 3-25 and. 3-26). Its annual cycle is consistent from
year-to-year. Elevated wintertime concentrations of particulate
sulfur in the air mass owe their existence to the relative
i
inefficiency of atmospheric pollutant scavenging processes
during winter. In arctic air masses, aerosols have a lifetime
of several weeks compared to. several days to a week at mid-lati-
tudes. In standard textbooks the atmospheric residence time of
SC>2 and sulfate are commonly given as about 1 and 3 days respec-
tively, which in the light of these recent findings can be mis-
leading. Furthermore, the longer -residence times in certain
regions and seasons help to explain why hemispheric background
sulfate, though less important than local or "long range transport"
effects, can still .be significant in producing acidic precipitation
episodes in remote areas, especially where the atmospheric con-
centrations of alkaline particles are small. The predominant
sources of particulates in the the North American Arctic are, in
order of decreasing importance, Siberia, Europe and eastern
North America.
There is evidence suggesting that the background concentrations
of sulfur oxides in a continental arctic air-mass that penetrates
the North America modeling area undergoes an annual cycle similar
to that observed in air nearer the pole. Weekly average and
weekly minimum concentrations of sulfur dioxide and particulate
sulfate at ELA-Kenora are shown in Figure 3-27 for the period
November 1978 to December 1979. The continental arctic air-mass
-------
3-48
Figure 3-25. Locations of Sites In the Arctic Aerosol Sampling
Network (M- Mould Bay, I- Igloolik and A- Alert)
and of the American Sampling Site (B- Barrow).
-------
3-49
3
E
O)
a.
CL
_J
D
to
CO
CO
LU
O
X
LU
0 L
I I I
I I I I
CLOUD COVER
MOULD
IGLOOLIK
1979
M~rA~T~M
1980
80
Qi
LU
8
U
0
O
g
u
Figure 3-26.
Weekly-Average Excess-Sulfate (non-sea salt)
Concentrations In the Atmosphere At Mould Bay
and Igloolik As Well As of Monthly Average Cloud
Cover In the Arctic. Source: Huschke, 1969.
-------
3-50
rvi
a .
LTI
2S
20
IS
10
1
1
1
1
1
1
1
1
1
1 - 1 - r
WEEKLY HVERHGE t MINIMUM RIR-CONCENTRHT IONS
LOCUTION; ELR-KENORP, PERIOD: NOVEMBER IBVB-DECEMBER 1979
n
2S
20
IS
II X
a
LTl
WEEKLY RVERBGE I MINIMUM RIR-CDNCENTRflTION5
LOCUTION: ELS-KENORH PERIOD: NOVEMBER ISVB-DECEMBER 1973
i r 1 _ i i - i i
NDJFMRMJJR 5 0 N D
Figure 3-27. Diagram of Weekly Average and Minimum SC>2 and
304 Concentrations At the Experimental Lakes
Site (see Figure 3-3) Between November 1978
and December 1979.
-------
3-51
is usually present at this sampling location between November and
April. SC>2 and sulfate background concentrations (as indicated
by the weekly minimum concentration) are a maximum between January
and March. They average 1 and 0.8 ;ag m~3, respectively. In
summer, they are below the detection limit (0.2 jjg m~3 S04=; 0.5
pg m S02).
The seasonal variation of precipitation pH in the polar air-
mass, as estimated by Barrie et al. (1981) from observed particu-
late acidity, is shown in Figure 3-28. The pH is expected to
vary annually from a CC>2 induced pH of 5.6 in summer to a late
winter/early spring pH of 5.0. This cycle is confirmed by measure-
ments of the pH of snow in a Ellesmere Island Ice Cap by Koerner
and Fisher (1981) shown in Figure 3-29.
There is some evidence that the background pH of snow in
northwestern Canada is depressed during winter. From a snowpack
chemistry survey around an isolated thermal generating station in
northwestern Alberta, Barrie (1980) found the background pH of
snow that accumulated between December 1977 and January 1978 to
be 4.9 to 5.
3.6 Conclusions
The Phase II Data Analysis review included annual,
seasonal, and episode deposition monitoring results and their
interpretation using emission inventories, ambient concentra-
tion data, and trajectory calculations. This review found
the highest precipitation acidity on an annual basis in the
-------
3-52
6.0
5.5
a
a
uj 5.0
4.5
JFMAMJ JASON D1 J 'F 'M A M J
1979 1980
Figure 3-28,
Acidity of Fresh Snowfall At Mould Bay Estimated
From Measured Sulfate Concentrations, Hydrogen
Ion To Sulfate Ratios In Aerosols and A Scavenging
Ratio of 2 x 105 (by volume).
-------
Figure 3-29.
5.5--
as
ex
5.0-
Seasonal Fluctuations Ihe pH of Snow In the
Agassiz Ice Cap Ellesmere Island For Each Year
Marked. Source: Koerner and Fisher, 1981.
TfT
62-
0
1 1 1 1 1 I 1
2
\ 1 1 1 r~
3
-r- i | i
4
i i
Depth of Ice (meters)
-------
3-54
Northern Hemisphere over eastern North America, Western
Europe, and Japan, with more alkaline precipitation over the
large continental areas of Western North America and Asia.
The cause of the slightly acidic precipitation along the west
coast of North America is not well understood, but could
be due to either anthropogenic sources which do exist or the
release of biologically-produced organic sulfur compounds
from the Pacific Ocean surface or both. The zone of maximum
acidity in eastern North America stretches in a corridor
through Ohio and Pennsylvania into Southern Ontario. Avail-
able concentration data at remote locations in eastern North
America generally indicate a summer sulfate maximum and a
winter S02 maximum with highly episodic behavior in pollutants
on a daily basis. In addition, calculated dry depositions
of sulfur are found to be of comparable magnitude to wet
sulfur depositions especially close to source regions and in
the winter season. Recent interpretations of both concentra-
tion and deposition monitoring data using trajectory calcula-
tions indicates that maritime tropical air masses from the U.S.
are the principal conveyors of elevated concentrations and
depositions in the extreme northeastern U.S. and southeastern
Canada, as opposed to continental polar air masses from Canada.
However, the source-receptor relationships based on back-trajec-
tory calculations and event data at single monitoring stations
are not always straight forward and contain some uncertainties.
-------
3-55
Some of the uncertainties are inherent such as trajectories in
precipitation systems and the non-linear chemistry, while others
can be reduced by analysis of data at more sites and for longer
periods. Finally, the data analysis review concluded that
within the Phase III modeling area of North America, natural
sources of sulfur are unimportant compared to anthropogenic
sources. In addition, the review found that background sulfur
concentrations enter the modeling area from the Pacific, the
Carribean and Atlantic Ocean south of 30°N, and the Arctic.
While the Pacific air masses are found to have concentrations
close to "hemispheric background levels" due to scavenging in
orographic precipitation, the Carribean and Arctic air masses
may contain somewhat elevated sulfur concentrations relative to
those in the modified Pacific air masses. The predominant sources
of elevated sulfur concentrations in North American Arctic air
masses in the winter are thought to be, in order of decreasing
importance, Siberia, Europe, and Eastern North America.
-------
Chapter 4
4. THE ROLE OF MODELING IN THE DEVELOPMENT OF EMISSION
CONTROL STRATEGIES AND AN AIR QUALITY AGREEMENT*
4.1 Introduction
A long-range transport model (LRT) is a description of
the physical processes involved in long-range transport in
the precise language of mathematics. Relationships between
components of the physical system are replaced with logical
connections or mathematical equations. Once a model has
been verified with monitored data it can be treated as a
"reasonable" analog of the real world. Then we can investigate
causal relationships between the variables of the physical
system by routine manipulation of the equations.
The system associated with long-range transport is
complex. The model describing this system consists of a
large number of submodels corresponding to the components of
the system. To keep the computing effort manageable, it is
usually necessary to keep the submodels of the LRT models as
simple as possible. This means that the long-range transport
model may not incorporate all our understanding of the relevant
physical processes.
In theory, we could construct a perfect LRT model if we
had all the information necesary to solve the mass, momentum
*See Addendum A for a personal critique of modeling by L.
Machta, U.S. Co-Chairman. See Addendum B for a response to
Addendum A by H. L. Ferguson, Canadian Co-Chairman.
-------
4-2
and energy conservation equations. In practice this is
impossible, and we have to be satisfied with using limited
information to estimate concentrations. This means that it
is difficult to predict the concentration field corresponding
to a given set of observations used to characterize the state
of the LRT system. However, we.can attempt to estimate
concentration field averages for the infinitely large ensemble
of possible concentration fields described by the set of
input observations such as velocity and precipitation. This
introduces the concept of an ensemble. Note that the ensemble
is defined by the input information used in the model to
estimate concentrations. A "good" model should be able to
predict the ensemble average. The ensemble is defined as the
average over a large number of individual model runs in which
only one or a few adjustable parameters are allowed to change.
It is important to reiterate that measurements of concentration
(or deposition) are expected to deviate from model predictions.
The concepts can be clarified through a simple analogy.
Consider an office with five rooms which can be occupied
by five people. There is no restriction on the number of
people who can occupy any one office at a given time and we
assume that the desire of any person to sit in any office
does not depend on the number of people already in the office.
In others words, the "concentration" of any one room can vary
from 0 person/room to 5 persons/room. Given this information
can we predict the occupancy of a specific room at any instant?
The best we can do is to say that the "average" occupancy of
-------
4-3
the room is 1, a number which is obtained by dividing the
number of occupants by the number of rooms. We know that this
ensemble average will differ from the room "concentration" at
any one instant. To understand this a litle better let us
compute the number of ways we can arrange 5 people among 5
rooms with no restriction on the number of people in any given
room. Using some basic combinational mathematics we can show
that this number is 126. Let us denote the number of combi-
nations with a specified number of people x in a given room
by N(x). Then we can show that N(0) = 56, N(l) = 35, N(2) =
20, N(3) = 10, N(4) = 4, N(5) = 1. These calculations indi-
cate that out of 126 possible combinations we are likely to
i
see 56 zeros, 35 ones, 20 twos, 10 threes and4 fours and 1
five. Note that the frequency of seeing a zero is the highest.
At any one instant the probability of seeing the ensemble mean
value of unity is only 28%. The mean value over a period of
time during which the occupants have gone through a large
number of. rearrangements is
C = 56x0 '+'35x1'+ 20x2 + 10x3 + 4x4 + 5x1 (4-1)
126
= 1.0
How long do we have to wait before the average concentra-
tion approaches the ensemble mean C. If the office is square
with sides of length L and u is the velocity with which the
occupants move around, the time taken for the rearrangement
-------
4"-4
of the offices is approximatley L/u. This suggests that we
have to wait several of these time scales for the average
measured concentration to approach the predicated -ensemble
mean. To make the discussion more quantitative let us calcu-
late the ensemble variance as follows
<(C -C)2> = [56x(0-l)2 + 35x(l-l)2 + 20x(2-l)2 +10x(3-l)2
. + 4x(4-l)2 +lx(5-l)2] 1 126
» l."33
Then, the expected deviation between measurement and prediction
(e2) as a function of averaging time: .
e2 = 1.33 L (4-2)
U T
If we take L = 15m and u = 0.25ms"1, the time scale governing
rearrangements of the office is approximately 1 minute. For e2
to become small, say 1/10, the averaging T should be approxi-
mately 15 minutes.
We can make our model a little less simple by using an obser-
vation of the number of occupants in one room. If j is the
observation, the prediction of the model is .
C = (5 - j)/4 (4-3)
If j = 1, C = 1 and if j = 2, Cf = 0.75. Note that the
ensemble mean prediction is a function of the way we define
an ensemble. Specifically, the ensemble is defined by the
value of the input 'j1 to the simple model. To consider the
implications of increasing the information content used in a
model consider the case when j = 1. We now have 4 people who
-------
4-5
can move around in 4 rooms and the number of ways this can be
done is 35. In terms of our previous notation we can write
N(0) =15, N(l) = 10, N(2) = 6, N(3) =3, N(4) = 1.
Note that the probability of observing the ensemble mean
increases slightly to 29%. The ensemble variance becomes
<(C - C)2> = 15x1 + 10x0 + 6x1 + 3x4 + 1x9 (4-4)
35
= 1.20
By using one observation, the variance of the prediction
decreases from 1.33 to 1.20. Note that the variance is also
a function of the definition of the ensemble.
If we refer to a model as complex when it uses more
observations, we see that the main difference between a simple
and a complex model lies in the expected deviation between
the model prediction and observation. We can decrease this
deviation by increasing the number of observations. However,
this will not hold true if the extra observations are in error
or if the complex model combines these observations in a
physically incorrect manner.
4.2 Uncertainties in Model Predictions
The relationship between a model prediction Cpj_ and
observation Coj_ can be expressed as:
Coi = Cpi + Ei (4-5)
since the observations are expected to be normally distributed
about the ensemble mean, a "good" model would be characterized by
E~i = 0 (4-6)
<3T2 = E.2 is small (4-7)
t i
-------
4-e
Note that Ej_ is a measure of the accuracy of the model and
is a measure of the precision of the model. Then, the
uncertainty in the model prediction is expressed in terms of
the statistics of the residuals between model estimates and
observations used to verify the model. The mean and variance
of E will not be useful if Coj_ is not normally distributed
about the ensemble mean. Empirical evidence suggests that
concentrations are lognormally distributed. This means that
residual statistics would be computed with log-transformed
concentrations. In the next Section, we will show how these
uncertainty estimates can be used in the process of decision
making.
4 .3 Use of Model Results in Decision Making
We have seen that model uncertainty can be expressed in
terms of residual statistics. To illustrate their use we
will consider two exmaples of decision making, hydrology and
construction.
Precipitation statistics are used routinely in the design
of dams and waterways (water control structures represent a
large investment activity on a continental, national or on state/
province scale). Similarly, climatological wind statistics may
be used for designing skyscrapers or building bridges.
Often the decision to build a bridge has already been taken;
the question is how it should be constructed to be safe. A
large safety factor is built in, even at significant extra
-------
4-7
cost, because the risks involved in not doing so are unaccept-
able. These characteristics are food for thought in the
acid rain control context. As in building bridges, where
the intangible of human life is at risk, the protection of
intangible environmental components is a major element to
consider.
In the case of regional average frost free period statis-
tics and similar data, economically significant decisions are
made regarding the choice of major agricultural crop options.
Here again, such data (constituting the general planning model)
proscribe the envelope of feasibility or the broad decision
framework. At a specific farm within the region, where more
detailed information is available or can be obtained, more
detailed plans can be developed. For example, it will be
recognized that the annual frost-free period varies spatially
as well as from one year to the next, and that a farm situated
in a valley or frost hollow can expect to have a frost free
period shorter than the regional norm. The planning strategy
to be used for that particular farm will represent a refine-
ment or subsequent level of planning within the general
regional "planning control parameters". Decisions leading
to the optimum management of the farm are not necessarily
going to be made all at once. The farmer may gather informa-
tion and experience over several years before achieving the
optimum. This doesn't mean that he doesn't do anything for
-------
4-8.
that period of time. In order to meet his particular objec-
tives he can implement a least-risk strategy based upon the
v*
information available.
The analogy between these climatic variables and current
concentration models (and deposition transfer matrices) is
also valid in another respect, in that the annual precipitation
over a region follows a log normal type distribution. Note
that one recognizes the possibility of an annual precipitation
value having a return period of 1000 years, but the building
of dams is not abandoned on the basis that construction for
flood control for the 1000-year storm is too expensive.
Instead, a more rational return period is selected and struc-
tures are installed on the basis that (a) flood control is
necessary and (b) there is a practical limitation to accep-
table' cost.
In summary, LRT regional long-term models have similarities
to simple climatic models in the following respects:
(1) They do not incorporate everything we know about
the physical problem (for reasons that can be
readily explained);
(2) They should not be applied to problems for which
they were not designed;
(3) They should not be dismissed or criticized on the
basis that they don't explain phenomena on finer
time and space scales than those for which they
are intended;
-------
4-9
(4) They can be, and in the case of climate models
have been, applied to major economic decisions in
many sectors; and
(5) They have similar statistical variabilities in
space and time which define significant (but
accepted) error limits when applied to individual
population elements.
These examples illustrate the general nature of decision
making. All decisions in the real world are made in the face
of uncertainty. We are uncertain about the course of future
events. Added to this is the uncertainty in our present
knowledge of the physical system we are interested in. We
cannot discount LRT models just because they provide uncertain
concentration estimates. We have shown that these predictions
can be useful if the decision maker explicitly accounts for
model uncertainty in making decisions.
4.4 Transfer Matrices
A transfer matrix is an array of numbers linking the
sources to the receptors linear. For example,
Ci = Tij E (4-8)
where Ej is the emission source j, C^ the concentration to
the receptor i and TJ_J is the transfer coefficient for
concentration.
Here, i = 1, 2, M and j = 1, 2, N, and a
summation is carried out over repeated indices. We can
-------
4-10
similarly formulate transfer matrices for deposition.
Transfer matrices can be generated by LRT models. Thus
the uncertainties in model outputs, described above, also
apply to the uncertainties in the elements of the transfer
matrix.
Appendix 6 gives an example of how transfer matrices
could be used for control strategy evaluation if the transfer
coefficients had no uncertainty. Since there is uncertainty,
the optimization must be based on probabilities. This type
of optimization procedure will be examined during Phase III.
-------
Chapter 5
5. SUMMARY OF SELECTED MODELS AND THEIR
INTERCOMPARISON/EVALUATION
5.1 Summary of Model Profiles
5.1.1 AES -'LRT ^
The three-dimensional trajectory model uses objectively
analyzed wind fields and computes vertical motions at four
pressure levels: 1000, 850, 700, and 500 mb. The analysis
procedure is essentially a three-dimensional scheme incorporat-
ing hydrostatic and height-wind balance routines (Rutherford,
1977), and producing gridded analyses of u and v wind components,
temperature, dew point and precipitation.. Input wind fields
are available every six hours and interpolation routines
are used to obtain winds at intermediate positions in time
and space. The computations are performed on the standard
Canadian Meterorological Centre grid of 381 km at 60° N,
with the capability of operating on sub-grid scales down to
95 km. Trajectory segment endpoints are determined each
time-step by assuming constant acceleration and using an
iterative scheme. The motion of air parcels can be followed
backward (receptor mode) or forward (source mode) from
anywhere in North America.
Trajectory paths are computed across grid cells of pollu-
tant emission, monthly mixing height and daily precipitation
amount. Uniform vertical mixing is assumed to occur instan-
taneously up to the mixing height, and transformation and
removal processes are linearly parameterized (Olson, et al.,
1979.
-------
5.1-2
The one-layer model parameterizes the physical and
chemical processes within a unit box extending vertically
from the ground to the inversion defining the mixing height.
*
Pollutant removal is parameterized by wet and dry deposition
and chemical transformation. Pollutant input to each box is
provided from an annual North American SC>2 or NC>2 emissions
inventory on a 127 km grid (Voldner and Shah, 1980). Instan-
taneous mixing occurs throughout the box.
The boxes follow trajectories that have been previously
computed and stored in 3-hour steps by the trajectory model.
At each step there is pollutant input from the inventory/
chemical transformation and surface deposition. The combina-
tion of these processes results in a new concentration value
within the box. The new concentration value is carried over
to the next point where the process is repeated.
The SC>2 to S0^= chemical transformation rate is assumed
to be constant. Dry deposition is parameterized in terms of
deposition velocities. Wet deposition is parameterized by
scavenging ratios and by a gridded daily array of precipitation
amount.
Trajectories from the AES trajectory model have been
compared to trajectories computed from various European and
American trajectory models (Olson, et al., 1978). Trajectories
computed at levels above the surface (i.e., 925 and 850 mb)
exhibit more agreement than surface trajectories.
-------
5.1-3
A numerical analysis of the trajectory model has been
conducted and a report is presently being prepared (Walmsley,
et al., 1981). An analytic non-divergent wind field was used
in the; intercomparison oK: model trajectory positions with analytic
solutions to the trajectory equation. No serious deficiencies
were found in the formulation of the model.
A short preliminary model evaluation (Olson, et al., 1979)
and a more complete model evaluation using EPRI-SURE data for
October 1977 (Voldner, et, al., 1980) have been reported.
The AES-LRT model is periodically evaluated with additional
network data; a summary of the current evaluation statistics
for the Phase I model for January and July of 1978 and for the
entire year are given in Section 5.2.
5.1.2 ASTRAP
The Advanced Statistical Trajectory Regional Air Pollution
(ASTRAP) model (Shannon, 1981) consists of three main subpro-
grams; vertical diffusion, horizontal dispersion, and calcula-
tion of surface concentrations and deposition. Other programs
for data preparation or presentation of output are not con-
sidered a part of the basic model, since such programs are
usually tailored to a particular application.
The vertical diffusion subprogram contains the dry
deposition and chemical transformation algorithms, as well
as a one-dimensional numerical solution of the standard dif-
fusion equation by the Gaussian Moment-Conservation technique
-------
5.1-4
(Shannon, 1979). Dry deposition is parameterized by deposition
velocities; the major difference between dry deposition in
ASTRAP and that in most other LRTAP models are that deposition
velocities in ASTRAP vary with time of day and season, and
that average SC>2 and 804 deposition velocities are almost
equal in magnitude.
The chemical transformation rate also has diurnal and
seasonal variations. There is an increased transformation
rate during initial dispersion (first three hours) for emis-
sions from the lower layer, in order to simulate the effects
of increased catalytic transformations in more polluted urban
areas.
Eddy diffusivity profiles which simulate the cycle of
nocturnal surface inversion formation, deepening, and erosion
from below, are specified by hour and season.
A diurnal variation in the emission rate is also included
in the vertical diffusion subprogram. The variation is a maxi-
mum in the surface layer, and decreases with height until
the variation about the average emission rate becomes zero
in the sixth layer (600-800 m). The variation of emission
rate is an arbitrary estimation of the effect of diurnal
variations of heating and cooling loads, working hours, and
the like.
-------
5.1-5
The number and thickness of the layers in ASTRAP is
optional. Since the eddy diffusitivity profiles and emission
inventories are specified for each layer/ they must be re-
i
estimated or relocated each time the layer definition is changed,
Vertical profiles of one-dimensional concentrations of
SC>2, primary 804, and secondary 804 (i.e./ that produced by
atmospheric transformation of SC>2) are calculated for norma-
lized emissions within each layer in turn. The advantage of
calculation in this structure is that scenarios of varying
fuel type/ or source region, or scenarios about different
stack heights, can be examined later without recalculation
of the profiles. .
The horizontal dispersion subprogram utilizes the concept
that long-term diffusion in the horizontal is determined by
\
the distribution of the plume centerlines rather than by
small-scale diffusion about the centerlines. The statistics
of endpoint locations are calculated and stored for six-hour
increments.
Wet removal is simulated by first advecting the tracer
for a six-hour time step and then checking the new location
to see whether precipitation has occured during the six-hour
period. If so, a fraction, F, of the sulfur mass represented
by the tracer is deposited as a function of the half power
of the precipitation amount (Hicks and Shannon, 1979). The
tracer portions deposited are stored by plume age for each
-------
5.1-6
source, and statistics similar to those for dry tracers are
generated. The fractional "dry" tracer remaining after a
precipitation event is (1-F) times the fraction at the
beginning of the event. The so-called dry tracer is the one
not subject to wet removal.
The statistics generated by the horizontal dispersion
program, for a grid of virtual sources and for each increment
of plume age, are the mean position and spread of the end
point ensembles and the number of equivalent dry tracers con-
tributing to the statistics, for both dry tracers and wet-
deposited tracers. Note that these statistics are independent
of sulfur species, because the wet removal applies to bulk
»
sulfur.
Statistics from the main subprograms are combined with
an emission inventory, including primary sulfate emission
factors, to produce output in a non-normalized form. The sub-
program combines the horizontal distribution statistics of dry
tracers with the normalized one-dimensional surface concentra-
tions and dry deposition increments; combines the horizontal
distribution statistics of wet tracers with the normalized
.one-dimensional budgets; and sums the resulting concentrations
and deposition from each source for a regularly spaced receptor
grid, or for a list of receptor locations.
-------
5.1-7
Basic products of ASTRAP simulations are regional long-
term average (monthly or longer) fields of S02 and 804 con-
centrations, and cumulative wet or dry deposition of total
sulfur. Dry deposition can he subcategorized as dry deposi-
tion of SC>2 and 804; however, the wet removal parametrization
is for bulk sulfur and thus speciation is artificial. By
integration of the deposition fields for specified subregions
of the grid (such as the eastern U.S. or eastern Canada)'
sulfur budgets are also obtained routinely. In general, ASTRAP
should only be applied in larger meso- to regional-scales and
monthly or seasonal time scales.
Adjunct subprograms of ASTRAP objectively analyze wind
»
and precipitation fields, grid emissions horizontally and
vertically, and contour simulated concentration and deposi-
tion fields on a background map of eastern North America.
The wind fields for ASTRAP are computed by first
calculating the mean wind between the surface and 1 km for
each rawinsonde observation, and then performing a form of
inverse distance-squared objective analysis at regularly
spaced grid points. The objective analysis scheme obviously
cannot improve upon the 12 hr and 200 km - 400 km resolution
of the raw data.
Hourly precipitation observations are summed for six
hours, combined with 6-hour precipitation observations, and
analyzed for a grid spacing of about 70 km. If no observation
site is inside a grid cell, the nearest observation is used.
-------
5.1-8
Emission data can be either a list of point sources,
with appropriate location and stack data, or a grid of virtual
sources, sorted by effective emission layer in either case.
5.1.3 ENAMAP
In the mid-1970's, SRI, International developed the
trajectory-type European Regional Model of Air Pollution
(EURMAP) for the Federal Environmental Office of the Federal
Republic of Germany (Johnson et al., 1978). In the late
1970's, the U.S. Environmental Protection Agency contracted
SRI, International to adapt and apply the EURMAP model to
eastern North America. The adapted model, ENAMAP (Eastern
North American Model of Air Pollution), is capable of calcu-
k
lating long-term SC>2 and 864 concentration and dry and wet
deposition patterns and regional and international exchanges
resulting from the emissions of S02 and 804. It should be
noted that another version of the model, ENAMAP-2, is currently
being produced. This second version will upgrade the para-
meterizations of vertical mixing, dry deposition, and trans-
formation and will include nitrogen chemistry by the end of
1982.
Basically, the ENAMAP model can be classified as a mass-
conserving Eulerian-puff model. Chemical processes within each
puff are parameterizated as they move across the modelling
domain. At 3-hour intervals, the concentrations and deposi-
tions are calculated from each puff and apportioned to the
grid cells on the basis of the portion of the puff within
each of the cells.
-------
5.1-9
Discrete puffs of S02 and 804 are released every 12 hours
from 80 km by 80 km emission grid cells. The mass of pollu-
tant in each puff is determined by dividing the annual emis-
sions by 730, the number of 12-hour periods in a year. These
puffs are tracked in 3-hour time steps until either they move
outside the model domain or their concentration drops to an
insignificant level.
The individual puffs are transported in a vertically-
averaged and horizontally-interpolated wind field. These
fields are updated every three hours.
Upon release, each puff is assumed to immediately diffuse
vertically to yield a uniform concentration in the layer be-
tween the surface and the mixing, height. In the model, the
mixing height varies seasonally, between 1150 and 1450 metres.
The constant transformation rate of S02 to S04= (1%/hr) in the
ENAMAP model was chosen after a review of field, laboratory, and
theoretical studies. The SC>2 and 804 dry deposition rates are
treated as constant throughout the simulation period. The dry
deposition rates, representing the daily average, are based
on reviews of fields, laboratory, and theoretical studies and
on an evaluation study conducted with the EURMAP model.
The wet deposition calculations are based upon hourly
precipitation rates and the duration of puff exposure to that
precipitation. The model does not distinguish between rain
and snow scavenging. Every three hours, the model determines
-------
5.1-10
via preprocessed, objectively-analyzed, three-hourly precipi-
tation fields the precipitation intensity in the vicinity of
the puff. Within each three-hourly simulation period, an
average hourly precipitation rate is estimated.
The monthly SC>2 and 804 deposition patterns are obtained
via summation of the contributions of each puff within each
receptor grid cell. The monthly-mean SC>2 and 804 concentra-
tion patterns are obtained in a similar manner.
Transport wind fields were generated at 3-hourly intervals
using the receptor grid network of 70-km spacing and the 12-
hourly surface and upper-air wind data available at approxi-
mately 60 sites in the United States. At each data site,
average u^ and \r components of the transport wind in the sur-
face layer were calculated. The transport winds were then
generated from these layer-averaged wind components by a
distance-weighted interpolation scheme using a weighting factor,
Precipitation fields were generated at 3-hourly intervals
for the same 70 km grid network.
Since the ENAMAP model was designed to consider primary
sulfate emissions, both S02 and 504 emissions were gridded
separately on an 80 by km UTM grid network. The annual
emission grids were generated by simply adding the annual
emission rates from all the point sources within each grid
square and the county-wide area sources of those counties
whose geographical center was locked in the grid square.
-------
5.1-11
No attempt was made to consider natural emissions of S02
or 804. Emissions crossing the western boundary and entering
the modelling domain also were not considered by the model.
5.1.4 OME-LRT
The OME-LRT model is statistical in that the physcial
processes of transport are expressed in terms of statistical
parameters. The basic premise of this class of models is
that long-term concentrations are insensitive to short-term
fluctuations in meteorology. It is assumed that concentra-
tions averaged over periods of the order of a year reflect
"mean" patterns of large scale meteorology. This allows one
to take a simple approach to the modelling of long-range
transport.
The model is based on the idea of classifying pollutant
particles as "wet" or "dry". Wet particles exist during
precipitation and dry particles during dry periods. Over a
long term, each travel time from a source is associated with
a particular mass of dry particles and a particular mass of
wet particles. Using this concept one can formulate differen-
tial equations for the evolution of these particles as func-
tions of travel time from the source. It is assumed that the
average rate of "conversion" from wet to dry particles is
inversely proportional to the average length of wet periods
in a Lagrangian sense. A similar assumption can be made
regarding "conversion" of dry particules to wet particles.
-------
5.1-12
Further, it is assumed that the scavenging coefficients do
not vary with travel time, but are different for wet and dry
periods.
The transformation of SC>2 to 804 is a complex process
that depends on a number of physical variables such as solar
intensity and ambient ozone concentration (see Wilson and
'Gillani, 1978). For long-term modelling we assume that the
conversion rate is 1%/hr, a value which is an "average" of
field measurements made during dry periods.
The dry deposition velocities chosen, 1.0 cm s"1 for S02
and 0.05 cm s"1 for S04, are based upon previous modeling
efforts and some field measurements. However, characteriza-
tion of dry deposition remains a contentious matter among field
experimentalists.
A table in the complete OME-LRT model profile presents
all the model parameters used in the simulations.
The long-term concentration C(t, t) at point r at time t
can be written as (Lamb, 1980)
_ t
C(r_, t) = Q r p(r, t|rs_, t1 )dt' (5-1)
oo
where Q is the emission rate of the source located at rs and
p(r> fcl£s' t') is the probability density that a particle
released at rs at time t1 will be found at r at time t. We
-o3 ^*^
assume that scavenging and dispersion are independent.
-------
5.1-13
The dispersion function will depend upon large scale
wind patterns. The parameters of the distribution x" (t),
y (tr) are the coordinates of the mean position of the
particles position after travel time 'tTfrom the release
point. The dispersion parameters C)x~ (t) and CJy('C) correspond
to standard deviations of particles about (3c,y) after travel
time "C from the source. These parameters can be determined
from trajectory statistics as suggested by Bolin and Persson
(1975).
The wet and dry deposition of sulfur depends on the
vertical distribution of the pollutant as well as the turbu-
lence in the diurnally varying planetary boundary layer.
The distribution of SC>2 and 864 was taken to be uniform
in the vertical through the depth of a constant mixed layer.
We should point out that this limits the resolution of the
model to distances of the order of 100 km from major sources.
The large scale horizontal distribution of pollutants
is determined by the parameters x/y/ <3x and <3y. For the
coordinates of the mean motion of large scale eddies we
assume that
x" = u tr
( 5-2)
y = 0
where u is the mean velocity of synoptic eddies and is the
travel time from the source.
The analysis of trajectories by Slinn et al (1979) and
Bolin and Persson (1979) suggests that Ox" and dy" can be
expressed as
-------
5.1-14
Oy =
(5-3)
On the basis of statistical dispersion theory it is
reasonable to assume that u and v are the standard deviations
of the horizontal velocity fluctuations of synoptic turbulence.
These statistics can be derived by sampling 850 mb winds over
periods of the order of years. Tennekes (1977) suggests
the following values for the large-scale velocities.
u = 10 ms'1, u = 10 ms"1, v = 6 ms"1
The emission data for the model was collected from
several sources. In the northeastern sector of the U.S.,
it was compiled from the EPA point source inventories
(Benkowitz; 1979) and the GCA (consulting company cotracted
by EPRI) major point and area source record. The Canadian
emission points except in Ontario were taken from the AES
preliminary point and area source inventory (Voldner et at.,
1980). The Ontario points were extracted from the Ontario
Ministry of the Environment sulfur dioxide emissions inventory.
All large (> 100 kTonnes/yr S02) point sources listed
in the above mentioned emissiond data and 95% of the major
V
(>10 kTonnes/yr S02) point sources were incorporated into
the model's inventory (usually grouped) to form effective
point sources located at the emissions-weighted geometric
means of the coordinates of the contributing points.
Approximately 60% of all area emissions and 72% of all minor
-------
5.1-15
point emisions were incorporated into the inventory by adding
minor point sources and area sources located near ( 50 km)
major points to that point or combining small sources concen-
trated in large urban centers to form effective point sources.
The model sensitivity to uncertainty in the value of the
input parameters was tested by independently varying each
parameter within the range of values cited in the recent
literature. For this test an idealized source-receptor
to the input parameters as a function of the source-receptor
orientation.
The interested reader is referred to the OME-LRT model
profile for the completed discussion.
5.1.5 RCDM-2
A simple approach which gives the same basic results
as the more computationally involved methods has been proposed
recently by Fay and Rosenzweig (1980). These authors have
assumed that the longer period sulfur dioxide and sulfate
concentrations from a point source can be described by the
steady state diffusion equation in which the horizontal eddy
diffusivity and conversion and removal rates are uniform in
space. Analytical solutions to the diffusion equations for
sulfur dioxide and sulfate concentrations are found under
these simplifying assumptions.
The sulfate predictions from the steady state model are
also in general agreement with those from the ASTRAP model.
RCDM-2 preserves the basic features that produce essentially
-------
5.1-16
the same mean transport field that one gets from a large
number of trajectories but eliminates most of the detailed
fluctuations. Seasonal and annual resultant wind vectors
at the upper air stations are used.
The two-dimensional steady-state advection-dif fusion
equation with removal is:
u ]*C + v 3C = Df }2C + "^2CN\ - (5-4)
where u and v are the mean wind velocities in the x and y
directions/ D^ is the diffusivity and f is the removal (wet
plus dry) time constant. This equation can be solved for
a steady point source at the origin having an emission rate
Q, with a boundary condition of. zero concentration at infinity;
C = Q expf w x*\ K frfl + fj2_Y\i/2 V (5-5)
^2^7 ' \ [_Dh-C \2Dh/ J J
where the x* axis is aligned in the direction of the mean wind.
velocity, w, Ko is the modified Bessel function of the zeroth
order, r is the radial distance from source to receptor, and
h is the mixing height. The completed deriviation of this
equation appears in RCDM profile report.
For a gridded emission inventory, the model assumes that
emissions are concentrated at the center of each square.
Because of certain model assumptions, there is no reason
to expect predicted concentrations in home grid squares to
be realistic, but they still must be finite, and still should
be consistent with model behavior in adjacent grid squares.
-------
5.1-17
For these reasons, the recent TRI work (Benkeley and Mills,
1980) computed home grid square concentrations as the area
integral (using Simpson's rule) along the mean wind axis in
the home grid square from its center to its edge. This
turns out to be equivalent to solving the equation at a
distance of 23.5 km. from the origin and this distance is
then used for all home square calculations.
The analytical solutions for the horizontal distribution
of primary and secondary pollutants from a steady point source
at the origin having an emission rate Q are:
C-, = _ Q_ exp/ w x'\ K \ r [~1 + /w \*\ V2 Y (5_6)
'
and
C2 = fiQ exp(_w_ xA K0(Yr) - KQ(^r)
h j oc.2 - ff2
(5-7)
respectively where
o,2 = /_!_ + _wL"\ (5-8)
V^°n ^l 1
x2 = / _i _ + -Sii-^ (5"9^
V-TTDh 4D^ )
Here the x' axis is aligned in the direction of the mean
wind velocity, w, Ko is the modified Bessel function of
zeroth order, r is the radial distance from source to receptor,
h is the height of the mixing layer, D is the horizontal
diffusivity, and p is the mass ratio of secondary pollutant
formed per mass of primary pollutant. The rate constant
-------
5.1-18
takes into account all forms of depletion of the primary
pollutant. The rate constant for loss of secondary pollutant is
" = ( T: g" + r') (5-10)
where the rate constants for wet and dry deposition of the
secondary pollutant are ""£~q and"C^.~ , respectively. Contri-
butions from multiple sources of primary pollutants can be
superimposed since the boundary condition at infinity has been
specified to have an ambient concentration of zero.
The Modeling Subgroup Report (section 2.5) and the RCDM
Model Profile contain more description of the RCDM model inputs,
sensitivity, and evaluation results.
5.1.6 CAPITA Monte Carlo
The Monte Carlo approach to simulation of physical and
chemical atmospheric processes has the following key characteris-
tics: (1) each process (transport, transformation, deposition)
is simulated directly as a discrete event; (2) mass conservation
is maintained by counting each pollutant quantum as it moves
and changes chemical form, rather than via differential
equations; (3) direct Monte Carlo simulation is Lagrangian
since trajectories of individual quanta are followed, but
Eulerian in that in the limit of an infinite number of quanta
and infinitesimal timestep a stochastic solution of a two
dimensional diffusion equation with non-steady state
inhomogeneous flow is obtained. Its Lagrangian features are
most useful for simulation of kinetics; the Eulerian aspects
benefit the transport simulation.
-------
5.1-19
The CAPITA Monte Carlo model represents an attempt to
provide a conceptually and computationally simple approach
for daily simulation of air pollutant concentration and
deposition on the regional scale. The model is continually
being modified to test alternative hypotheses; the discussion
below describes a currently operational version which is not
expected- to change dramatically in the near future.
Emissions are represented by virtual point sources/
representing arbitrary mixtures of actual point and area
sources. The emission grid spacing is nominally 190 km,
except in the northeastern U.S. and southeast Canada, where
the grid spacing is 95 km. The emission grid, on a polar
stereographic projection true at 60°N, does not include
natural sources nor western North America sources.
The advection of pollutants is facilitated using surface
data. The rationale for the use of surface winds instead of
the more commonly used upper air winds is given by Patterson,
et al., 1981.
The data for all reporting sites for each parameter were
interpolated onto a square grid via a weighting function.
Each gridpoint weighs about 3 to 6 of the nearest National
Weaher Service sites. A further smoothing is applied to the
grid. The wind vector components northward and eastward were
gridded and smoothed separately.
The general features of the wind vector field from upper
air and surface wind data are similar. However, it is evident
-------
5.1-20
that the higher station density network of surface data yields
more structure in the wind field.
Surface wind speed is always less than the wind
speed for the bulk of the planetary boundary layer. A
calibration factor is required therefore by which the surface
wind speed is increased to match the mean "mixed layer" speed.
This factor is obtained by taking the ratio of the properly
averaged upper air wind speed to the surface wind speed for
every grid point. The seasonal variation of the scale height
over which the upper air winds are averaged were taken from a
study by the Atmospheric Environment Service of Canada
(Portelli, 1977).
The concept of "the trajectory" of an air parcel during
multiday transport has no physical realization. After repeated .
subjection to diurnal planetary boundary layer dynamics of
turbulent diffusion, oscillations in mixing height, and
vertical wind shear, a puff of emitted material may be dispersed
beyond recognition into part of a large scale "background" airmass.
For computational simplicity, the "diffusion" is currently
simulated by an effective diffusion coefficient, K, such that
a displacement of radius /2KAt is imposed with randomly
chosen direction following the advection step. In the limit
of large numbers of quanta, the Monte Carlo approach is both
Lagrangian and Eulerian. In the model simulation, only one
trajectory is chosen, so that the single quantum emitted at a
-------
5.1-21
source in the mid-West might be exported to the Great Lakes
within 3-4 days or remain in the Rocky Mountain states after
10 days, depending on the sequence of random diffusion steps
it experiences.
The general features of the simulated spatial distribution
are preserved without distinct plumes if a value of K = 105m2s~l
is assumed, but K = I06m2s~l causes an overly uniform simulated
spatial distribution. We have adopted a value K = 4 x lO^m^s"-'-.
Just as for the random walk nature of diffusion, the
CAPITA Monte Carlo approach treats the kinetics of chemical
transformation, dry deposition and wet removal as stochastic
events. The model requires specification of transition
probabilities. Over each timestep, all possible paths (S02
conversion to 804, etc.) are assigned a probability of
occurrence. Equivalent simulation of the ensemble mean is
achieved by algebraic allocation of mass among the various
forms, treating the transition probabilities as expected
fractions of conversion and removal.
Currently, the model assumes that S02 emissions contain
1% primary S04=, and lumps wet and dry deposition together.
The values used are constant diurnally.
For the Phase II effort, the SC>2 deposition was presumed
to be 95% dry deposition and 5% wet. Sulfate was allocated
such that 20% was dry and 80% of 804 deposition was via
"precipitation".
-------
5.1-22
Simulation .of wet removal will be included in Phase III
of the MOI activity. The formulation requires knowledge of
the probability of the pollutant experiencing precipitation
within a timestep. From the spatially dense hourly precipitation
data, the number of reports of precipitation divided by number
of reporting sites yields the likelihood of rain. Removal
percentage for both SC>2 and 804, given that the pollutant is in
a precipitating airmass, then multiplies the probability of
precipitation to yield the desired transition probabilities
of wet removal.
^ The CAPITA Monte Carlo model has been compared to sulfur
concentration and deposition data for both episode studies
(Patterson et al., 1981) and seasonal averages. Earlier
studies indicated that the transport so dominates that the
simulation is somewhat insensitive to both source specification
and kinetics for episode studies.
The model evaluation and sensitivity studies, as adopted
by the Modeling Subgroup, are presently being performed for
the CAPITA model.
5.1.7 MEP-TRANS
A regional trajectory based model, TRANS (Transport of
Regional Anthropogenic Nitrogen and Sulfur) was developed to
satisfy both requirements of resolving the patterns of single
plumes near the sources and the effects of large scale atmos-
pheric motions in transporting the plumes over distances of
one thousand kilometers or more.
-------
5. 1-23
The horizontal wind field is obtained from the sea-level
pressure field distribution through the so-called geostrophic
approximation with some adjustment being made in both speed
and direction in order to be representative of the 300 to 500
metre level above the surface.
The observational data consists of sea-level pressures
obtained at 6-hourly intervals on the CMC 381 km grid.
A method due to Sykes and Hatton (1976) is used in which
orthogonal polynomials in the two space co-ordinate x and y,
and in time co-ordinate t, are fitted to the data. The space
polynomials are defined on the network of observing points by
recurrence relations. The series of six-hourly observations
for the day allow the determination of pressure at any time
between observations. The spatial pressure field is fitted
within one millbar error corresponding to a wind field error
of the order of one m/s.
Chung (1977) has derived a method by which a wind field
representative of the atmospheric motions at a height of
300 to 500 meters above the ground may be obtained from the
geostrophic wind. In this approach used in TRANS, both
speed and direction adjustments are obtained by means of a
regression equation in terms of the geostrophic wind speed
and the 3-hour pressure tendency.
The trajectory integration is done by a method due to
Peterssen (1956), in which an iterative procedure is used to
determine the new position after travel time At.
-------
5.1-24
The integration can be carried forward in time to deter-
mine the motion of the air parcel from an arbitrary location,
and can be carried backward in time to determine the history
of air parcels arriving at an arbitrary location.
The accuracy of the integration procedure.was determined
from tests carried out over 96 hours of travel time in both
forward and backward trajectory modes with a 3-hour timestep.
It showed a maximum difference in spatial position of 1 km,
indicating a negligible source of error in comparison with
the uncertainties in the pressure fields.
The pressure field uncertainties are due to fitting errors
and errors in the original data. Only a small fraction of the
observational data error (about 0.5 mb) is being transmitted to
the interpolated values for the missing data. The fitting
procedure introduces an error in the pressure field but tends
to eliminate random noise inherent in the original data. Frontal
features are usually not resolved by the pressure data, leading
to error in the determination of the trajectories near fronts.
Typically, for randomly oriented velocity errors, trajectory
positional error may be of the order of 50 km after 1 day's
travel. Where a sharp gradient in velocity exists, as in a
strong depression, the position of an air parcel after 24
hours travel can be several hundred kilometers from the true
position.
The pollutant material contained in the air parcel
centered on the plume trajectory is assumed to disperse with
-------
5.1-25
time, so that the horizontal distribution about the centerline
can be approximated by a normal (Gaussian) distribution. The
vertical distribution is assumed to be uniform.
The mixing height is allowed to undergo a standard
diurnal cycle ranging from 0.2 to 2.0 times the seasonal mean
value in four discrete vertical steps. The pollutant material
in each of the four layers is transported by the same wind field
and participates in the wet deposition. Only the pollutant
material below the mixing height is assumed to be subject to
dry deposition.
The deposition of material from the plume onto the ground
is modeled by means of the deposition velocity concept.
Within the model/ the deposition velocity is dependent
on time but is not explic.ity made variable in the spatial
coordinates.
Washout or rainout is parameterized by means of a bulk
washout parameter.
The washout coefficient is proportional to precipitation
rate and is specified for each plume segment separately.
Precipitation rate for a given segment is determined from 3-
hourly precipitation intensities over the network of observing
stations. Thus the washout of material is governed by the
precipitation encountered by each plume segment in its travel
path as determined by historical record.
For the case of SC>2 washout, the approach suggested by
Barrie (1981) has been adopted. This approach allows the
-------
5.1-26
temperature and pH dependence of the SC>2 washout ratio to be
explicitly incorporated. The chemical transformation of 302
to 304 in the plume is modeled as a first order reaction.
The implementation of the transformation in the model is through
modification of the source terms after each time step.
It is found that for normal conditions, the decay time
of SO2 is about 24 hours and total sulfur, 45 hours. Similarly,
noontime emission produces a S02 decay of"time of about 36
hours and total sulfur decay of time 72 hours. Similar
computations for NC>2 and total nitrogen, give 16 hours and 72
hours respectively, for standard conditions. -
A high transformation rate k of S02 to 804 of 0.025 per
hour generates and maintains 10 times as much 304 in the
plume as does the low rate of 0.001 per hour. The total
sulfur in the plume at long travel time is twice as high for
the high transformation rate, due to the lower sulfate
deposition velocity.
It is found that the effect of the deposition velocity of
304 on sulfate levels indicates that only a 50% variation is
evident after 24 hours for deposition velocities in the range
of 0.1 to 2.0 cm/s.
An effective mixing height of 1200 metres leaves twice
as much SC>2 and nearly 50% more sulfate in the plume after 24
hours as does the 500 m mixing height.
-------
5.1-27
Moderate rain (5 mm hr~-M produces a very rapid removal
by plume washout/ the S02 decay time decreases to 9 hours
with what little sulfate forming initially being washed out
very rapidly.
The nitrogen behavior was found to be similar to the
sulfur behavior described above.
The emissions data for SC>2 and NOX used in the modeling
study were the 127 km griddded data currently being used by
AES in their regional transport evaluations.
A total of 70 trajectory origins were selected to coincide
with grids having a large -emission rate. For each trajectory
origin, 5-day trajectories at 3-hourly time steps were
generated for the full year. The trajectories are used in
conjunction with the 3-hourly precipitation data to define
the precipitation intensity for each segment of the plume.
Concentrations and loadings were calculated at 75 receptor
points. The receptor locations were chosen to coincide with
the 11 sensitive receptor areas chosen by the working group
and a number of monitoring stations in the APN, CANSAP, MAP3S,
NASP, EPRI, NADP and Ontario Hydro networks to facilitate
subsequent comparison with data. The values of the parameters
for the 1978 comparison where selected based on the sensitivity
tests.
The MEP model annual S02 concentration ranged from
1 ug/m^ in Northern Ontario and Northern Quebec to 30 ug/m^
-------
5..1-2 8
over a wide region of the Eastern U.S./ with an isolated peak
of 50 ug/m3 in the Ohio Valley region. The annual average
SO^ concentrations ranged from 1 ug/nr in Northern Ontario and
Quebec to 20 ug/m3 in Eastern U.S. with a peak of 25 ug/m3 in
the Ohio Valley region. The N02 concentrations range from
0.5 to 20 ug/m3 with two regions of higher concentration
in the Detroit and New York areas. Predicted NO^ ranged
from 1 ug/m3 in Northern Ontario to 20 ug/m3 in the New
York area.
Maximum dry deposition of S02 of the order of 25 kg.S.ha."1
yr."1 occurred in association with the high concentration
areas. The 804 dry deposition was approximately 10% of the
S02 deposition. Dry deposition of nitrogen was up to 5 kg.N.ha
yr. for N02 and 1 kg.N.ha.^yr. for NOj".
The predicted wet deposition of S02 ranged from 0.5
kg.S.ha.~lyr.-1 in Northern Ontario and Quebec to 15 kg.S.ha."1
yr."1 in Ohio with an isolated peak of 20 kg.S.ha.^yr.~1. The
wet deposition of SO^ reached 10 kg.S.ha.^yr. in the Ohio
region. Wet deposition of nitrogen ranged from 0.1 to 5 kg.N.ha."1
yr."1, being approximately equal for the two species.
Wet deposition of sulfur ranged from 1 kg.S .ha.^yr .-1 in
Northern Ontario and Quebec to a peak of 30 kg.S.ha.'iyr."1 in
Ohio. Wet depositions of nitrogen ranged .from 0.5 to 10
kg.N.ha.^yr."1 in the lower Great Lakes regions.
-------
5.1-29
Total (wet plus dry) sulfur deposition showed a large
area of excess of 30 kg.S.ha.~lyr."-^ centered in the
Ohio Valley and extending into Southern Ontario. Total
nitrogen depositions expressed as N are approximately one-
third of the sulfur depositions on a regional basis.
5.1.8 UMACID
The Atmospheric Contribution to Inter-regional Deposition
(ACID) model is a receptor-oriented model designed to estimate
V
the contributions of upwind sources to measured pollutants at a
given receptor. The model incorporates the horizontal diffusion
due to vertical wind velocity shear, chemical transformation,
spatially and temporally varying dry deposition and "events"
precipitation scavenging.
The model is intended for use at rural locations where
local sources are negligible.
The use of a mixed layer trajectory model assumes that
the material being traced is moving with the mean motion of
the mixed-layer. This assumption will only be true if the
material is, in fact, well-mixed through the layer and there
is no shear in the layer to disperse the material away from
the mean flow.
Over sufficiently long travel time (>1.3 hours), the
dispersion of atmospheric admixtures will be dominated by
vertical wind velocity shear. The climatology of this
-------
5.1-30
dispersion for long travel times has been calculated by
Samson (1980) from the divergence of trajectories for sublayers
of the mixed-layer. The distribution of the probability of
contributing to a measured concentration due to a given mixed
layer trajectory was shown to be normally distributed about
the mixed-layer. The broadening of the probability field
with length of travel was found to be a function of time.
The ACID model used this parameterization to describe
the potential contribution of upstream sources ignoring
intermediate chemical transformation, dry deposition and wet
deposition. The dispersion due to a meander of trajectory
centerlines in times is inherently calculated by sequential
integration of trajectories at six hour intervals over the
period of interest.
Following the work of Shieh et al., (1979), ACID includes
spatially varying dry deposition with changes in dry deposition
fields from season to season. To simulate the variation in
dry deposition due to changes in atmospheric stability, the
gridded deposition velocities are forced to vary diurnally.
The reaction rates associated with the chemical trans-
formation of pollutants will change the probability of being
affected by an upwind source. Ignoring deposition, a source
close to a receptor may have less chance of producing an
impact on secondary pollutant levels than a source several
hours upstream because of the insufficient time available for
-------
5.-1-31
transformation to occur. ACID assumes a diurnally varying
transformation rate.
The approach used for wet deposition in the ACID model
is a first order loss rate. By use of a scavenging coefficient,
the change of probability of contribution is of a simple form.
ACID allows parts of puffs to be washed out without affecting
the remaining region of the puff. This is accomplished
through a unique technique involving a moving grid following
the course of the trajectory. An initial probability of
contribution field is calculated at the first time step
upstream (t = 3 hrs) based on dispersion alone. At each fixed
grid point within the domain of contribution the change in
probability due to dry deposition and chemistry is applied.
Next/ the gridded precipitation field is scanned and the
probability reduced for precipitation areas.
The location of the fixed grid within the domain is
memorized on a "moving" grid which is moved upstream to the
second time step. The probabilities are reduced on the moving
grid due to chemical conversion. .Then the moving grid is
translated to the underlying fixed grid through an efficient
Gaussian smoothing and interpolation scheme whose response
has been designed to simulate the dispersion due to vertical
shear.
The new fixed grid is again subjected to dry deposition
and wet deposition and the probability is reduced accordingly
-------
5. 2-1
and added to the fixed grid. The location of these fixed
points are then memorized and moved to the third time step.
The total process is then repeated through all time steps.
The sum of the probabilities can be further summed~over
defined regions and used to define the transfer matrix of
source-to-receptors contributions.
5.2 Intercqmparison and Evaluation
There is no general agreement in the modeling community
as to the proper method and statistics for intercomparison
and validation of models. However, some desirable conditions
of the intercomparison process are that input meteorological
and emission data should be common, when possible, and that .
evaluation statistics should be based upon residuals between
simulations and observations, where available.
Since both meteorological process and emission rates
vary by season, it is instructive to examine simulations for
months and seasons, as well as simulations of annual concen-
trations and deposition. However, the ability to treat temporal
variations varies among models.
Model simulations have inherent minimum resolutions, since
there are limitations on input data, as well as internal modeling
limitations. Observations, on the other hand, are point values.
Since the observations are random realizations of some ensemble
expectation, perfect model/observation correlations can never be
obtained.
-------
5.2-2
The models described in 5.1 have been developed for
different purposes, and no single model is likely to be optimal
for all applications. The relative value of the models for
particular applications has yet to be determined.
A preliminary comparison of model estimated average
concentrations and wet depositions has been completed for the
AES-LRT, ASTRAP, MOE-LRT, and RCDM models.
Comparison of AES, ASTRAP and RCDM monthly (January and
July) average SC>2 and 804 concentrations and total wet deposit-
ions with monitoring data gave the following results:
For January 1978, the AES-LRT on average over-predicted
both the S02 and SC>4 concentrations. The ASTRAP model SC>2
concentrations were on average closer to the observed SC>2
concentrations. These comments are based on only 21 SC>2 and
26 804 data points; hence, statistically there is no discernible
difference between the performance of these two models. The
RCDM model almost consistently under-predicts SC>2 concentrations
for January 1978 by a factor of 2 or greater; however, January
304 concentration estimates are on the average very close to the
observed concentrations. The wet sulfur depositions from AES-LRT
and ASTRAP were on the average significantly larger than the
observed depositions. Predictions of January wet sulfur deposit-
ion by RCDM are on the average close to the observed values.
However, no overall conclusion can be drawn due to the small
small number of observations (11 stations) and some uncertainty
in the deposition data.
-------
5.2-3
The July 1978 model SC>2 and 804 concentrations are on the
average not as close to the observed as the January estimates
were. Both AES-LRT and ASTRAP over-predicted the S02 concen-
trations; however, on the average, the ASTRAP simulations for
the 20 sites are close to the observed. The converse is true
for 804 concentrations (30 stations) in that both models on the
average under-predicted, with AES-LRT simulations closer to the
observed. RCDM model simulations of July S02 concentrations are
on the average very close to the observed concentrations; how-
ever the RCDM model underestimates the 804 concentrations for
the same period. July wet sulfur depositions were on the aver-
age slightly over-predicted by the AES-LRT and ASTRAP models and
were on the average under-predicted by the RCDM model. The
standard deviations for the three models are large and there
were only 15 observations.
The comparison of the annual model averages with observed
data was possible for only a small data set. Based on 9 stations
the observed annual SC>2 concentrations on average were over-est-
imated by ASTRAP, AES-LRT and MOE-LRT. The RCDM model under-est-
imates on the average. For MOE-LRT and ASTRAP the over-prediction
was less than by the AES-LRT; however, statistically there was no
discernible difference due to the small sample size. The annual
804 concentrations are on the average under-estimated by the MOE-
LRT and ASTRAP models at the 13 sites considered. (Note that the
ASTRAP annual average is based on only the arithmetic average of
-------
5.2-4
2 monthly averages, namely, January and July). The AES-LRT model
on the average slightly over-predicts the 804 concentrations. The
simulations of the RCDM model were on the average very close to
the observed concentrations.
Annual wet sulfur depositions were slightly over-predicted
by all four models when assessed on an average of the 8 CANSAP
data sites reporting at least 10 months of data in 1978. The
deviations about the geometric mean"were large for both the ASTRAP
AES-LRT models; however, the MOE-LRT showed much smaller deviat-
ions between the observed and the predicted. The RCDM predictions
were on the average much lower than the-observed depositions.
During Phase II a hierachy of model evaluation statistics
were adopted, but only the AES-LRT model computed all of them
(see the AES-LRT Model Profile Report and section 4.2.2 of the
Modeling Subgroup Report). It is expected that these statistics
will be available from all eight models by September 1981, in an
Addendum to Chapter of the Model Subgroup Interim "Working Re-
port".
-------
Chapter 6
PHASE'II'IMPROVED'EMISSION'INVENTORY
6.1 Emissions
To eliminate the differences in the emission inventories
used .in the Phase I modeling work, a U.S^-Canada SC>2 emission
inventory was produced bilaterally during Phase II in coopera-
tion with Work Group 3B. However, the Modeling Subgroup
found it difficult to standardize the source areas in Phase
II because of the re-programming required by some of the
models. The receptor areas were standardized to an extent
as will be described in the next section. The inventory
was produced on a state/province, 63 ARMS area and top 50
point source .basis to accommodate the needs of all 5 Phase I
models.
The best estimates of the current U.S.-Canada SC>2
emissions at the State and Province and the ARMS area levels
are presented in Tables 6-1 and in Table 6-2, respectively.
A list of the top 50 point sources in the eastern U.S. and
Canada is provided in Table 6-3. The background material to
these data has been incorporated in the Phase II SC"2 Emission
Inventory Report (No. 2-4). This report supercedes the
information in Appendix 6 and the Addendum to Appendix 6 of
the Phase I report of Work Group 2.
Work Group 2 is continuing, to work with Work Group 3B to
produce best estimates for primary sulfate, NOX, and historical
-------
6-2
Table 6-1. Phase II Improved United States and Canadian S02 Emissions
On a State and Province Basis (Kilotonnes/Yr) - 1980
State or Total
Province
Alabama 603.6
Arkansas 100.4
Connecticut 118.3
Delaware 121.8
Dist. of
Columbia 36.1
Florida 839.1
Georgia 580.1
Illinois 1,259.5
Indiana 1,968.0
Iowa 290.0
Kentucky 1,178.3
Louisiana 253.1
Maine 95.9
Maryland 232.0
Mass. 334.6
Michigan 939.6
Minnesota 384.4
Mississippi 205.1
Missouri 1,286.0
New Hamsphire 59.9.
New Jersey 416.4
New York 860.0
North
Carolina 447.9
Ohio 2,907.0
Penn. 1,643.5
Rhode Island 26.3
South
Carolina 260.2
Tennessee 1,034.0
Vermont 13.3
Virginia 386.5
West
Virginia 1,144.9
Wisconsin 604.4
SUB TOTAL 20,612.2
Newfoundland 56.5
Prince Edward
Island 15.8
Nova Scotia 217.3
New Brunswick 223.8
Quebec 1,105.3
Ontario 1,842.8
Manitoba 515.2
SUB TOTAL 3,976.7
Utilities Non-Ferrous
Smelters
443.9
52.9
12.2
87.1
_*
650.5
479.0
892.0
1,475.1
188.7
1,065.0
21.5
21.5
135.1
151.1
674.9
213.7 67.5
147.1
1,096.4 58.4
33.9
187.9
324.3
282.2
2,331.9
986.9 35.3
1.8
173.1
895.8 12.9
0.0
238.2
1,012.0
449.6
14,725.3 174.1"
5.3
5.9
88.2
91.7 12.0
1.6 609.6
436.9 929.8
7.4 485.5
637.0 2,036.9
Transportation
10.3
6.5
7.8
1.3
22.6
16.5
27.6
17.9
9.2
9.4
7.6
2.0
7.5
1.6
13.1
23.0
10.4
5.8
14.1
1.6
19.5
31.4
13.9
35.0
32.0
2.4
7.7
13.9
1.1
14.4
4.9
11.6
4"DTTT"
2.0
0.4
3.1
2.9
18.3
21.3
4.6
52.6
Ind/Res/
Co mm.
95.7
26.8
97.3
20.2 .
10.6
91.2
51.2
250.5
382.4
82.6
80.6
54.3
57.2
76.5
167.6
205.0
75.1
23.7
101.0
22.1
148.1
466.4
109.2
433.1
392.6
21.6
69.2
93.4
11.9
113.2
102.1
131.8
4,'OS87T
41.9
9.5
108.9
100.5
386.2
260.0
13.3
920.3
Other
53.7
14.2
1.0
13.2
2.9
80.9
22.3
99.1
101.3
9.3
25.1
157.3
9.7
24.8
2.8
36.7
17.7
28.5
16.1
2.3
60.9
37.9
42.6
107.0
196.7
0.5
10.2
18.0
0.3
20.7
25.9
11.4
1 , 2 51 . d
7.3
17.1
16.7
89.6
194.8
4.4
329.9
* Emissions included with Maryland
Source of U.S. Data: Mitre Corporation, April 24, 1981, and E.H. Pechan and
Associates, May 30, 1981
Source of Canadian Data: Environment Canada (Frank Vena)
-------
6-3
Table 6-2. Phase II United States and Canadian 502 Emissions
For the 63 ARMS Areas (Kilotonnes/Yr) - 1980.
ARMS
AREA
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
UNITED-STATES CANADA
82.6 7.5
. 5.8 12.4
6.7
52.4
355.0
11.2
. 74.1
9.4
220.7
343.3 36.5
611.9
452.1
390.0
675.2
30.1
266.3
125.1
353.6
975.7
148.5
465.0
723.5
675.6
532.3
28.9
274.9
165.9
243.3
429.7
130.2
530.3
ARMS
AREA
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
UNITED-STATES
150.0
27.6
130.2
583.5
200.9
312.3
91.7
3.1
1,289.8
291.5
721.8
402.2
645.5
1,446.0
2,006.1
544.4
339.7
981.8
48.8
611.8
314.8
CANADA
48.4 .
3.8
0.5
1.2
370.6
872.5
1.3
628.5
18.1
332.7
831.5
5.5
171.9
10.9
Source of U.S. data: Mitre Corporation, April 24, 1981, and E. H. Pechan
& Associates, May 14, 1981
Source of Canadian data: Environment Canada (Frank Vena)
-------
Table 6-3. Combined U.S.-Canadian Top 50 Sources of SC>2 Emissions - 1980
Rank
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Plant
Name
Inco
Noranda
Paradise
Inco
Muskingum River
Gavin
Cumberland
Clifty Creek
Baldwin
Monroe
Labadie
Kyger Creek
Harrison
Johnsonville
Mitchell
Hatfield Ferry
Eastlake
Bowen
Lambton G.S.
Gibson
Nanticoke G.S.
HBMS
Conesville
Shawnee
Algoma Steel
Bruner Land
State/
Province
Ontario
Qjebec
Kentucky
Manitoba
Ohio
Ohio
Indiana
Indiana
Illinois
Michigan
Missouri
Ohio
West Virginia
Tennessee
West Virginia
Pennsylvania
Ohio
Georgia
Ontario
Indiana
Ontario
Manitoba
Ohio
Kentucky
Ontario
Pennsylvania
Emission
(kilotonnes/yr )
807.5
537.5
418.8
333.5
306.7
297.5
296.2
295.3
237.2
224.3
222.6
219.7
215.1
188.0
187.3
173.5
172.8
170.3
160.3
187.8
155.1
152.0
151.8
146.1
143.3
139.3
Rank
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
i
Plant
Name
Montrose
New Madrid
Sammis
Coffeen
Krammer
Big Bend
Falconbridge Nickel
Keystone
Petersburgh
Conemaugh
Widows Creek
Mount Storm
Cardinal
Stuart
Joppa
Thomas HL
Montour
Kincaid
Gallatin
Gallagher
Gaston
Kingston
Avon Lake
Lake view G.S.
State/
Province.
Missouri
Missouri
Ohio
Illinois
West Virginia
Florida
Ontario
Pennsylvania
Indiana
Pennsylvania
Alabama
West Virginia
Ohio
Ohio
Illinois
Missouri
Pennsylvania
Illinois
Tennessee
Indiana
Alabama
Tennessee
Ohio
Ontario
Emission
(kilotonnes/yr )
138.0
135.1
133.7
124.8
123.8
122.7
122.3
121.5
120.9
118.2
118.8
116.4
115.0
113.1
107.6
101.8
96.9
96.0
95.6
94.6
93.6
92.0
91.7
91.4
Source of U.S. data: EPA Airtest Program (1979-1980)
Source of Canadian data: Environment Canada (Frank Vena)
-------
6-5.
SOX and NOX emissions on a state/province basis. In addition,
best estimates for SOX and NOX emissions in the western states
and provinces will be developed in Phase III.
6.2 Source-Receptor Areas
To generate the Phase II tranfer matrices and evaluate
the models, the subgroup decided to adopt a composite of the
Phase I U.S. and Canadian targeted sensitive areas, (see
Table 6-4). This provided the two "receptor oriented"
models with the minimum number of Phase II receptor areas.
The other 6 Phase II models ran on either the 11 Canadian
source regions and 9 receptor areas or the full 63 ARMS
areas (all these areas were considered potential receptor
areas as well as source areas).
In the latter part of Phase II, it was decided that the
emission source regions to be used in the development of ah
MOI transfer matrix in Phase III be resolved spatially on
a province-subprovince/state-multi-state level, as shown in
Figure 6-1. Eastern Canada has been subdivided into
15 source areas and the Eastern U.S. into 25 source areas.
These areas are also to serve as receptor areas in Phase III.
The emission and geographic centroids in longitude and latitude
of the Phase III areas are presented in Tables 6-5 and 6-6.
-------
6-6
Table 6-4. Phase II Targeted Sensitive Areas
For Work Group 2 Modeling
Area Longitude and SURE Grid
Number Name La t i t ude (approx*) Ce n t r oids (X,Y)
1 Boundary Waters 93°, 49° 4, 26
2 Algoma 84°, 46.5° 12, 22
3 Muskoka 79.5°, 45° 17.5, 19.5
4 Quebec City 72°, 47° 12.5, 22.5
5 Southern Nova Scotia 66°, 44° 30.5, 19.5
6 New Hampshire 72°, 45° 25, 20
7 Adirondacks 74°, 44° 23.5, 18.5
8 Western Pennsylvania 78°, 41° 18, '13
9 Southern Appalachia 84°, 35° 12.5, 5.5
10 Arkansas 92°, 36° 2.5, 7.5
11 Florida 82°, 30° 13.5, -1.5
*may vary by +0.5°
-------
6-1. Canadian Province and Sub-provin<^Hkegions and U.S.
State and Multi-state Regions for^rie Phase III
Transfer Matrices
en
i
VERMONT
NEW HAMPSHIRE
MASSACHUSETTS
RHODE ISLAND
CONNECTICUT
NEW JERSEY
DELAWARE
MARYLAND
DISTRICT OF COLUMBIA
-------
6-8
Ible 6-5. Canadian Regions For Phase III Transfer Matrices
Region
10
11
12
13
14
15
16
77
18
19
20
21
22
23
24
,-,,!,,, 1 . 1 ,,111
Province/
Sub-Province
Northern Manitoba
Southern Manitoba
Northern Ontario
West
Northern Ontario
East and
Algoma South
Sudbury
Southwestern
Ontario
Southeastern
Ontario
St. Lawrence
Valley - Quebec
Noranda and
' North-Central
Quebec
Gaspe Bay-Quebec
New Brunswick
Nova Scotia and
Prince Edward
Island
Newfoundland and
Labrador
Saskatchewan and
Alberta
British Comumbia
'
Area
Lat.
56.84
51.21
52.02
49.03
46.32
43.56
45.23
46.88
54.28
50.02
46.21
45.46
51.43
50.00*
49.30*
Centroids
Long.
97.77
98.34
89.20
83.03
80.70
81.24
77.40
72.13
73.59
64.56
66.77
63.27
58.89
110.00*
123.00*
Emission
Lat.
55.61
49.97
49.32
48.05
-
43.39
45.07
46.24
50.20
49.35
45.75
45.52
49.12
50.00*
49.30*
Centroids
Long.
98.25
98.72
90.00
83.76
-
80.59
75.38
72.70
74.02
66.14
65.98
62. 57
56.08
110.00*
123.00*
lay be revised
-------
6-9
ble 6-6. U.S. Regions For Phase III Transfer Matrices
Emissions Centroids Area Centroids
R,eg.ipn
50
51
52
53
54
55
56
57
58
59
r
61
62
63
64
65
66
67
68
69
70
71
72
1,
s.ta.te
Ohio
Illinois
Pennsylvania
Indiana
Kentucky
Michigan
Tennessee
Missouri
West Virginia
New York
Alabama
Wisconsin
Iowa
Minnesota
Virginia
North Carolina
Florida
Georgia
South Carolina
Maryland
Delaware
New Jersey
Dist. of Col.
Arkansas
,LpAg,«- , La,t,., ,;.,.,
82.5
88.0
79.0
87.0
85.5
84.0
86.0
91.5
80.5
74.0
87.0
88.5
92.0
93.5
77.8
79.0
82.0
84.0
80.8
76.3
75.3
74.4
77.0
92.5
40.0
41.5
40.8
40.0
37.5
43.0
35.9
38.8
38.5
42.0
34.0
43.5
42.0
45.0
37.8
36.0
29.0
33.5
33.8
39.2
39.1
40.5
38.9
34.8
Lpng...
82.5
89.0
77.7
86.1
85.5
84.5
86.0
92.5
80.5
76.0
86.8
89.7
93.5
94.5
78.8
79.0
82.0
83.5
80.8
76.3
75.3
74.5
77.0
92.5
,Lat,.
40.2
40.0
41.0
40.0
37.5
44.5
35.9
38.5
38.5
42.7
33.0
44.5
42.0
46.2
37.8
35.5
29.0
33.0
33.8
39.0
39.1
40.2
38.9
34.8
-------
6-10
Table 6-6 (continued).
Emissions Centroids Area Centroids
BBK 'j Vi
74
75
76
77
78
79
80
81
82
83
84
i
^^^
86
87
88
89
90
91
92
93
94
95
96
i S,t.a.t.e
Louisiana
Mississippi
Massachusetts
Connecticut
Rhode Island
Maine
Vermont
New Hampshire
North Dakota
South Dakota
Nebraska
Kansas
Oklahoma
Texas
Montana
Wyoming
Colorado
New Mexico
Idaho
Utah
Arizona
Washington
Oregon
Nevada
California
jlofig.' Lat. r' '' JLojrcg. Lat.' '
91.0 . 30.5 92.0 31.0
89.5 33.0 89.5 32.5
71.5 41.7 72.0 41.7
72.7 41.5 72.7 41.7
71.5 41.7 71.6 41.7
69.0 44.5 69.0 45.5
73.0 44.0 72.7 43.9
71.5 44.0 71.5 44.0
99.0 47.0 100.3 47.0
100.0 44.5 100.0 44.5
97.0 41.0 100.0 41.5
97.0 38.5 98.3 38.5
96.8 36.0 97.8 35.5
97.0 31.0 99.0 32.0
Note: The appropriate centroids for states 39-49 will be determined
in Phase III
-------
Chapter 7
7. SOURCE-RECEPTOR RELATIONSHIPS
"FOR"SULFUR OXIDES
7.1 Introduction
Several long-range transport models are currently available
for estimating annual sulfur deposition and for studying source-
receptor relationships. These models ae briefly described in
Chapter 5 and more completely in Chapter 2 of the Modeling Subgroup
Report. Some preliminary attempts to model nitrogen oxides are
discussed in Chapter 9.
Eastern North America has been subdivided in several ways for
studying source-receptor relationships as described in Chapter 4 of
the Atmospheric Modeling (Work Group 2) Phase I Report. In Phase
II, all models retained their original format for emissions data
input and source-receptor transfer matrix elements were determined
for the Canadian 11 source regions and 9 Phase I targeted sensitive
areas on an annual basis (see Figure 7-1 and Table 7-1). The models
used at least January and July 1978 meteorological data and the
Phase II sulfur dioxide emissions inventory in Chapter 6 of this
report.
7.2 Source-Receptor Matrix Description
In both Phases I and Phase II, annual S02 and 804 concen-
tration, and dry, wet and total sulfur desposition transfer matrices
were generated by the MOE, AES, ASTRAP, ENAMAP, and RCDM models.
-------
7-2
Figure 7-1. Map of East-
ern North America Showing
the 11 Major Source Regions
(10 shaded and the rest
east of the Mississippi
River) and 9 Sensitive
Areas Used in the Phase II
Transfer Matrices
-------
7-a
Table 7-1. Key to 11 Source Regions and 9 Sensitive Areas
Source Regions
1 Michigan 1
2 Illinois - Indiana 2
3 Ohio 3
4 Pennsylvania 4
5 New York to Maine 5
6 Kentucky-Tennessee 6
7 West Virginia to North Carolina 7
8 Rest of Eastern U.S. (Florida 8
9 Ontario 9
10 Quebec
11 Atlantic Provinces
Sensitive Areas
Boundary Waters
Algoma
Muskoka
Quebec
Southern Nova Scotia
Vermont-New Hampshire
Adirondacks
Western Pennsylvania
Smokies
-------
7-4
Since in Phase I, the ENAMAP Model domain did not inrlndp th^
Atlantic provinces of Canada, the effects of emissions from this
region were not considered by the model.
In Phase I, the modelers did not use the same meteorological
periods and emissions inventory. In order to provide a means
whereby the matrix element values computed by each of the
models could be intercompared, the values were normalized using a
unit emissions rate of 1 Tg of sulfur per year. The normalized
values for annual wet sulfur deposition (kg.S.ha.~lyr.~1) generated
during Phase I are presented in Table 7-2. The actual sulfur
emission rates used for each region by each model are presented
in the third column of this table.
The matrix element with the greatest variation among models
is that for the impact of the Ontario source region (9) on Muskoka
(3). The values ranged from 1.60 (MOE) to 12.86 (ENAMAP) kg.S.
ha.'^yr."1 per Teragram of sulfur emission. However, the
large value calculated by the ENAMAP model was probably due to
the location of the Muskoka receptor area about 250 km closer to
the Sudbury, Ontario, point source than in the other models.
Other matrix elements where the values varied significantly
were the impact of the Pennsylvania source region (4) on the
Pennsylvania sensitive area (8) [4.24(RCDM) to 10.61 (ENAMAP)];
-------
Table 7-2. Phase I Transfer Matrix of:
Annual Wet Deposition of Sulfur (kg.ha~l.yr~l)
per unit emission (Tg.S.yr~l)
(1)
Source
Regions
1
Mich.
2
111.
Ind.
3
Ohio
4
Penn.
5
N. York
to Maine
6
Kent.
Term.
Models
MOE
AES
ASTRAP
ENAMAP
RCDM
MOE
AES
ASTRAP
ENAMAP
RCDM
MOE
AES
ASTRAP
ENAMAP
RCDM
MOE
AES
ASTRAP
ENAMAP
RCDM
MOE
AES
ASTRAP
ENAMAP
RCDM
MOE
AES
ASTRAP
ENAMAP
RCDM
Erniss.
(Tg.S)
0.784
0.973
1.194
1.194
1.194
2.538
1.937
2.077
2.077
2.077
1.983
2.381
2.163
2.163
2.163
1.021
1.028
0.990
0.990
0.990
1.143
1.204
1.208
1.208
1.208
1.202
1.418
1.473
1.473
1.473
B. Waters
(1)
0.07
0.21
0.44
0
1.11
0.06
0.05
0.24
0
0.79
0.04
0
0.08
0
0.30
0.03
0
0.01
0
0.10
0.02
0
0
0
0.05
0.03
0
0.07
0
0.28
Alg.
(2)
0.40
2.40
2.59
0.97
3.95
0.23
1.20
1.32
0.11
1.14
0.15
0.25
0.86
0.06
0.80
0.12
0.29
0.29
0
0.36
0.07
0.17
0.08
0
0.21
O.lO
0.14
0.45
0.01
0.38
Musk.
(3)
0.93
3.20
2.50
1.67
2.51
0.32
1.10
1.40
0.08
1.35
0.32
1.80
1.84
0.36
2.32
0.28
1.30
1.13
0.04
1.62
0.19
0.50
0.44
0.01
1.08
0.14
0.71
0.76
0.01
0.63
, Recei
Que.
(4)
0.34
1.00
1.46
0.13
1.68
0.15
0.31
0.75
0.02
0.81
0.19
0.46
1.04
0.02
1.23
0.21
0.68
1.10
0
1.06
0.25
1.30
0.79
0
0.99
0.09
0.07
0.42
0
0'.36
5tor Areas
S. N.Sc.
. (5) .
0.39
0.31
0.55
0.19
0.38
0.18
0.10
0.37
0.02
0.26
0.28
0.21
0.60
0.09
0.52
0.40
0.29
1.20
0.07
0.95
1.00
2.00
2.51
2.86
2.33
0.13
0.07
0.24
0.01
0.17
Vt. NH.
(6) .
0.56
0.72
0.73
0.20
0.75. .,
0.23
0.30
0.52
0
0.48
0.32
1.00
0.91
0.05
0.96 .
0.39
1.80
1.79.
0.07
1.59
0.56
2.20
3.00
1.28
2.98
0.14
0.21
0.38
0.01
0.29
Adir.
(7).
0.86
1.10
1.14
0.11
1.11
0.31
0.36
0.82
0.01
0.74
0.47
1.30
1.50
0.12
1.60
0.57
2.20
2.59
0.32
2.63
0.80
2.40
2.50
0.97
3.19
0.18
0.42
0.63
0.01
0.46
Penn.
(8) .
1.70
1.70
0.81
0.13
1.09
0.76
1.10
1.18
0.06
1.40
2.00
4.70
3.29
1.01
3.66
4.40
7.90
4.61
10.61
4.24
0.33
0.42
0.69
0.07
0.88
0.46
1.50
1.88
0.13
1.48
Smokies
(9)
0.12
0.21
0.07
0.14
0.32
0.47
0.77
0.60
0.36
0.91
0.23
0.25
0.53
0.09
0.83
0.11
0.10
0.07
0
0.43
0.05
0
0.01
0
. 0.13
1.60
3.10
4.22
4.24
3.95
01
-------
Table 7-2. Phase I Transfer Matrix of: (continued)
Annual Wet Deposition of Sulfur (kg.ha
per unit emission (Tg.S.yr~l)
7
W.Virg.
to N.C.
8
Rest of
(USA) Fid
to Mo. to
Minn.
9
Ontario
10
Quebec
'"11
Atlantic
Provinces
MOE
AES
ASTRAP
ENAMAP
RCDM
MOE
AES
ASTRAP
ENAMAP
RCDM
MOE
AES
ASTRAP
ENAMAP
RCDM
MOE
AES
ASTRAP
ENAMAP
RCDM
MOE
AES
ASTRAP
ENAMAP
RCDM
1.703
1.223
1.610
1.610
1.610
1.196
3.743
4.012
4.012
4.012
0.906
0.985
0.949
0.949
0.949
0.595
0.519
0.464
0.464
0.464
0.187
0.235
0.453
0.453
0.453
0.03
0
0.02
0
0.14
0.09
0.24
1.09
1.90
2.16
0.08
0.10
0.09
0
0.29
0.06
0
0.02
0
0.02
0.01
0
0
*
0
0.08
0
0.34
0.01
0.35
0.39
0.61
0.56
0.46
0.92
0.51
1.80
2.92
1.32
2.04
0.18
0.19
0.52
0
0.14
0.03
0
0.02
0.15
0.33
0.80
0.11
1.08
0.34
0.24
0.44
0.15
0.42
1.60
3.30
4.71
12.86
5.07
0.32
0.58
1.14
0
0.47
0.05
0
0.10
0.02 | 0.06
0.13
0.33
0.61
0
0.65
0.15
0.05
0.29
0.03
0.32
1.00
1.70
3.83
0.59
4.10
1.50
2.90
2.61
0
1.23
0.16
0.43
0.63
0.11
0.28
0.25
0.56
0.02
0.52
0.15
0.03
0.15
0.03
0.08 . .
0.57
0.61
1.17
0.20
0.86
0.73
0.96
2.06
2.20
1.50 .
0.74
2.60
3.07
2.42
0.22
0.90
0.87
0.05
0.79
0.20
0.08
0.20
0.02
0.15 . .
1.10
1.60
1.41
0.16
1.74
2.30
3.30
2.09
1.50
2.41 .
0.16
0
2.48
0.60
0.29
1.10
1.24
0.06
1.23
0.26
0.13
0.27
0.02
0.23
1.20
2.00
3.08
0.63
2.51 ,
0.59
1.50
1.90
0.04
1.33
0.10
0
1.23
0.23
0.85
3.50
4.04
2.67
4.99. .
0.40
0.53
0.50
0.31
0.46
0.53
1.20
0.43
0.18
0.90. .
0.13
0.19
0.12
0
. 0.15
0.05
0
0.08
0.04
0.16
0.49
1.18
0.26
1.39
1.00
2.50
1.40
0.14
1.21 .
0.05
0
0.02
0.01
0.14
0.03
0
0
0
0.02
0.01
0
0
0.01
cr\
ENAMAP did not consider source region 11 in the Phase I modeling
(1) Annual deposition transfer matrix elements for the ASTRAP and ENAMAP models are based on an
arithmetric mean of the equivalent January 1978 and July 1978 elements.
-------
7-7
Michigan (1) on Algoma (2) [0.40 (OME) to 3.95 (RCDM)]; Ontario
(9) on Quebec (4) [0.59 (ENAMAP) to 4.10 (RCDM)]; Ohio (3) on
Pennsylvania (8) [1.01 (ENAMAP) to 4.70 (AES)]; and Virginias and
North Carolina (7) on Pennsylvania (8) [0.85 (MOE) to 4.99(RCDM)].
In Phase II, several changes in the input data and the para-
meterizations in some of the models resulted in revisions of the
values of the transfer matrix elements. These changes included
the use of:
1) the same meteorological periods (wind and
precipitation data for January and July 1978),
2) the same sulfur emissions inventory (1 Tg sulfur
from each source region)
3) the same locations of receptor areas (after
Phase I, it was discovered that not all the
modelers were using the same location for several
receptor areas)
4) a more complex and realistic wet and dry sulfur
deposition parameterization scheme in the
ENAMAP model.
5) a change in the relative percentage of wet and dry
sulfur deposition in the RCDM model
With the use of standardized input data, one would expect the
variations in each set of transfer matrix values to decrease
somewhat.
-------
7-8
Table 7-3 presents the normalized (1 Tg sulfur per year)
Phase II annual wet sulfur deposition transfer matrix values.
The complete set of Phase II transfer matrices are presented
in Appendix C of the Phase II Modeling Subgroup report.
7.3 Intercomparison of Phase I and Phase II Transfer Matrices
In this section, selected Phase I and Phase II annual
wet sulfur deposition transfer matrix elementvalues are inter-
compared. In addition, the 1978 wet sulfur deposition calculated
by the models at each targeted receptor area are compared to
measured values.
Table 7-4 presents selected transfer matrix element values
computed in Phase I and Phase II. The values in these elements
varied the greatest in Phase I.
For three of these six source-receptor pairs (Ontario-Muskoka,
Pennsylvania-Pennsylvania, and Virginias and North Carolina-
Pennsylvania) the maximum value and the range of values of wet
sulfur depositon increased. The range of values in the other
three source-receptor pairs decreased. The maximum value decreased
markedly in Phase II for the Michigan-Algoma (from 3.95 to 2.40)
and Ontario-Quebec (from 4.10 to 2.16) source-receptor pairs.
The deposition parameterization in both the RCDM and ENAMAP
models were modified significantly in Phase II. The ASTRAP model
was not modified, but it was rerun using the standardized meteoro-
logical and emissions data sets.
-------
Table
Phase II Transfer Matrix of:
Annual Wet Deposition of Sulfur (kg.ha~l,
per unit emission (Tg.S.yr~l)
Source
Regions
1
Mich.
2
111.
Ind.
3
Ohio
4
Perm.
5
N. York
to Maine
Models
MOE
AES
ASTRAP
ENAMAP
RCDM
MEP
MCARLO
MOE
AES
ASTRAP
ENAMAP
RCDM
MEP
MCARLO
MOE
AES
ASTRAP
ENAMAP
RCDM
MEP
MCARLO
MOE
AES
ASTRAP
ENAMAP
RCDM
MEP
MCARLO
MOE
AES
ASTRAP
ENAMAP
RCDM
MEP
MCARLO
Emiss.
(Tg.S)
0.784
0.973
1.194
1.194
1.194
2.538
1.937
2.077
2.077
2.077
1.983
2.381
2.163
2.163
2.163
1.021
1.028
0.990
0.990
0.990
1.143
1.204
1.208
1.208
1.208
Receptor Areas
B. Waters
(1)
0.07
0.21
0.06
0.04
0.30
0.14
0.06
0.06
0.05
0.07
0.00
0.25
0.05
0.01
0.04
0.00
0.02
0.00
0.11
0.01
0.00
0.03
0.00
0.00
0.00
0.04
0.01
0.00
0.02
0.00
0.00
0.00
0.02
0.00
0.00
Alg.
(2)
0.43
2.36
1.47
0.71
0.94
1.24
0.50
0.23
1.24
0.61
0.11
0.37
0.32
0.23
0.16
0.25
0.31
0.00
0.31
0.14
0.08
0.13
0.29
0.14
0.00
0.16
0.08
0.03
0.08
0.17
0.02
0.01
0.10
0.05
0.01
(1) Annual deposition matrix elements for AST
Musk.
(3)
0.92
3.19
3.44
1.40
1.37
1.27
0.61
0.31
1.14
1.02
0.58
0.50
0.31
0.28
0.32
1.85
0.75
0.02
0.91
0.95
0.33
0.29
1.26
0.48
0.00
0.84
0.60
0.20
0.22
0.50
0.12
0.04
0.53
0.66
0.17
Que.
(4)
0.33
1.03
1.10
1.32
0.73
0.18
0.33
0.15
0.31
0.71
0.59
0.28
0.08
0.20
0.18
0.46
0.92
0.75
0.47
0.23
0.33
0.21
0.68
0.74
0.73
0.54
0.45 '
0.34
0.26
1.33
1.38
0.60
0.47
1.24
0.35
S. N.Sc.
(5)
0.37
0.31
0.20
0.00
0.28
0.08
0.06
0.18
0.10
0.12
0.00
0.12
0.02
0.04
0.28
0.21
0.25
0.00
0.23
0.08
0.05
0.39
0.29
0.52
0.00
0.37
0.20
0.05
0.98
1.99
0.59
0.00
0.53
1.06
0.08
Vt. NH.
(6)
0.54
0.72
1.03
0.73
0.35
0.25
0.31
0.22
0.26
0.70
0.46
0.17 .
0.12
0.17
0.31
1.01
0.83
0.99
0.38
0.33
0.27
0.38
1.75
1.94
1.55
0.79
0.83
0.36
0.59
2.24
3.02
1.63
1.30
2.29
0.52
Adir.
(7)
0.83
1.13
2.33
1.35
0.57
0.42
0.40
0.30
0.36
1.16
0.35
0.28
0.13
0.28
0.46
1.34
1.79
0.90
0.65
0.47
0.47
0.57
2.24
2.15
3.91
1.37
1.28
0.50
0.83
2.41
2.94
3.46
1.71
1.89
0.29
Penn.
(8)
1.66
1.75
1.08
1.56
0.69
0.66
0.2*7
0.74
1.14
1.04
1.12
0.66
0.17
0.32
1.99
4.75
3.91
7.24
1.88
1.97
0.61
5.42
7.88
12.16
12.82
1.37
6.46
0.68
0.40
0.42
0.31
3.37
0.46
0.63
0.22
Smokies
(9)
0.12
0.10
0.07
0.79
0.18
0.11
0.20
0.47
0.77
0.40
0.56
0.54
0.38
0.43
0.24
0.25
0.24
0.23
0.42
0.29
0.23
0.12
0.00
0.04
0.24
0.13
0.16
0.05
0.05
0.00
0.01
0.06
0.06
0.01
0.01
[RAP and ENAMAP models are based on the mean of the
equivalent January 1978 and July 1978 elements.
-------
Table 7-3. Phase II Transfer Matrix of: (continued)
Annual Wet Deposition of Sulfur (kg.ha~-"-.yr~l]
per unit emission (Tg.S.yr~l)
6
Kent.
Tenn.
7
W.Virg.
to N.C.
8
Rest of
(USA) Fid
to Mo. to
Minn.
9
Ontario
10
Quebec
11
Atlantic
Provinces
MOE
AES
ASTRAP
ENAMAP
RCDM
MEP
MCARLO
MOE
AES
ASTRAP
ENAMAP
RCDM
MEP
MCARLO
MOE
AES
ASTRAP
ENAMAP
RCDM
MEP
MCARLO
MOE
AES
ASTRAP
ENAMAP
RCDM
MEP
MCARLO
MOE
AES
ASTRAP
ENAMAP
RCDM
MEP
MCARLO
MOE
AES
ASTRAP
ENAMAP
RCDM
MEP
MCARLO
1.202
1.418
1.473
1.473
1.473
1.703
1.223
1.610
1.610
1.610
1.196
3.743
4.012
4.012
4.012
0.906
0.985
0.949
0.949
0.949
0.595
0.519
0.464
0.464
0.464
0.187
0.235
0.453
0.453
0.453
0.03
0.00
0.03
0.00
0.14
0.01
0.00
0.03
0.00
0.00
0.00
0.06
0.01
0.05
0.09
0.24
0.24
0.12
0.37
0.23
0.06
0.08
0.10
0.04
0.02
0.17
0.07
0.20
0.06
0.00
0.00
0.00
0.03
0.03
0.01
0.01
0.00
0.00
0.00
0.00
0.00
0.00
0.10
0.14
0.22
0.00
0.18
0.05
0.04
0.08
0.00
0.05
0.00
0.15
0.02
0.01
0.38
0.61
0.38
0.15
0.25
0.34
0.11
0.62
1.83
5.43
1.45
1.30
2.17
0.47
0.20
0.19
0.19
0.00
0.12
0.34
0.09
0.03
0.00
0.00
0.00
0.01
0.01
0.00
0.14
0.71
0.40
0.03
0.28
0.09
0.10
0.16
0.33
0.25
0.00
0.51
0.16
0.09
0.33
0.27
0.37
0.02
0.18
0.12
0.09
1.63
3.35
6.89
1.41
1.91
2.81
0.61
0.38
0.58
0.65
0.00
0.30
0.49
0.28
0.06
0.00
0.00
0.00
0.04
0.06
0.01
0.09
0.07
0.14
0.05
0.16
0.03
0.11
0.13
0.33
0.37
0.38
0.30
0.09
0.14
0.15
0.05
0.12
0.08
0.12
0.02 '
0.05
1.00
1.73
2.15
1.61
1.52
0.72
0.44
1.58
2.89
7.71
.0.90
0.59
3.53
0.56
0.18
0.43
0.02
0.07
0.10
0.30
0.05
0.13
0.07
0.04
0.00
0.07
0.01
0.01
0.27
0.25
0.26
0.00
0.18
0.05
0.02
0.15
0.03
0.03
0.00
0.05
0.01
0.01
0.55
0.61
0.16
0.00
0.56
0.14
0.08
0.71
0.96
0.25
0.00
2.33
0.57
0.21
0.75
2.55
0.87
0.00
0.93
1.42
0.17
0.13
0.21
0.52
0.12
0.11 .
0.04
0.07
0.22
0.90
0.84
0.83
0.37
0.22
0.15
0.20
0.08
0.15
0.12
0.07
0.03
0.05
1.04
1.62
1.18
1.20
0.64
0.89
0.37
2.99
3.28
'1.54
3.06
0.81
3.15
0.20
0.19
0.00
0.03
5.78
0.33
0.19
0.04
0.17
0.42
0.96
0.17
0.19
0.06
0.16
0.29
1.14
1.03
1.77
0.60
0.28
0.20
0.25
0.13
0.26
0.15
0.11
0.05
0.09
1.14
2.03
2.18
1.75
0.95
1.56
0.32
0.65
1.54
2.11
1.35
0.53
1.87
0.17
0.12
0.00
0.00
5.39
0.12
0.10
0.01
0.45
1.48
1.30
1.16
0.61
0.18
0.26
0.91
3.52
5.98
7.73
2.79
1.25
0.42
0.39
0.53
0.32
0.19
0.25
0.05
0.08
0.54
1.22
0.25
0.53
0.39
0.99
0.17
0.13
0.19
0.12
0.03
0.13
0.16
0.10
0.05
0.00
0.00
0.35
0.02
0.07
0.00
1.64
3.10
8.17
3.91
1.29
2.01
0.75
0.18
0.49
0.13
2.73
0.48
0.75
0.17
0.99
2.46
2.74
3.05
0.81
0.29
0.20
0.05
0 00
0.02
0.25
0.06
0.05
0.09
0.03
0.00
0.00
0.03
0.03
0.00
0.01
0.02
0.00
0.00
0.09
0.01
0.00
0.00
H-1
o
-------
7-11
Table 7-4. Selected Transfer Matrix Element Values of Annual
Wet Sulfur Deposition from Phase I and Phase II
Source-Receptor
Range of
Phase I Values
(kg.S.ha.~!yr.~1/Tg.S.yr.
Range of
Phase II Values
(kg.S.ha.'lyr.'VTg.S.yr."1)
Ontario-Muskoka
Pennsylvania-
Pennsylvania
Michigan-Algoma
Ohio-Pennsylvania
Virginias and
North- Carolina-
Pennsylvania
Ontario-Quebec
1.60 - 5.07
4.24 - 10.61
0.40 - 3.95
1.01 - 4.70
0.85 - 4.99
0.59 - 4.10
1.60 - 6.90
1.37 - 12.17
0.40 - 2.40
1.88 - 4.70
0.85 - 5.98
0.72 - 2.16
-------
7-12
The wet sulfur deposition transfer matrix element values
generated by the RCDM model decreased by an order of magnitude.
The largest RCDM and ASTRAP Phase I values and the corresponding
Phase II values are "listed in Table 7-5.
Each of the RCDM Phase II values decreased markedly to 1/2
l/3rd their Phase I value). On the other hand, the ASTRAP Phase
II values tended to increase from their Phase I values, due
primarily to the difference in meteorological periods. (i.e. the
increase from 4.61 to 12.17 in the Pennsylvania - Pennsylvania
relationship.
7.4 Comparison of Model Estimates with Observations
Individual transfer matrix elements cannot be directly verified,
but total wet deposition can be compared to observations. Some
observations of wet sulfur deposition have been reported and have
been compared with model estimates at these sites (see Phase I Report)
Table 7-6 is an enlargement of that data comparison showing the same
five Phase I and Phase II models.
Although the Phase I model estimates show a wide variation at
some of the sensitive areas, they all show a maximum value at the
Pennsylvania site and have an average value of about 25.0 kg.S.ha."!
yr.'1 and a standard deviation of 6.4 kg.S.ha.-1yr."!. The average
value overestimates by about 30% because of the tendency of some
models to overestimate by 30% because of the tendency of some
-------
7-13
Table 7-5.
Significant Changes in Phase I and II
RCDM and ASTRAP Annual Wet Sulfur Depositions
Model
RCDM
ASTRAP
Source- receptor
Ontario-Muskoka
Virginias and NC-Penn
Ontario-Quebec
Michigan-Algoraa
Ky/Tenn- Smokies
Ohio-Penn
N. York to Maine-
Adir.
Ontario-Muskoka
Penn-Penn
Ky/Tenn-Smokies
Virginias and NC-Penn
Ontario-Quebec
Ohio-Penn
Ontario-Adir
Ontario-Algoma
Phase I Wet
Sulfur Deposition
(kg ha'1 yr -1 )
5.07
4.24
4.10
3.95
3.95
3.66
3.19
4.71
4.61
4.22
4.04
3.83
3.29
3.08
2.92
Phase II Wet
Sulfur Deposition
t (kg ha -1 yr'1)
2.79
1.91
1.52
0.94
1.29
1.88
1.71
6.90
12.17
8.17
5.98
2.16
3.92
2.19
5.43
-------
Table 7-6. Preliminary Model Estimates and Observations of Annual Wet Sulfur Deposition
(kg.S.ha.~^yr~l) at the Nine Targeted Sensitive Areas
Phase I Phase II
Sensitive Obs.l1'
Areas Values
1 B. Waters 3
2
3
4
5
6
7
8
9
Algoma
Muskoka
Quebec
S.N. Scotia "
Vt., N.H.
Adirondacks
Pennsylvania
Smokies
3
6
5
3
7
7
12
7
1 Canadian
OME*
3
5
7
6
7
8
8
17
7
AES
2
10
18
9
6
13
16
34
17
United States
RCDM
16
20
23
17
10
15
19
27
18
ASTRAP
7
16
21
17
12
15
19
25
17
ENAMAP
10
6
19
1
11
4
3
21
8
Canadian
OME*
3
5
7
6
7
8
8
17
7
AES
2 .
10
18
9
6
13
16
34
17
MEP
1
6
10
6
3
8
9
16
7
United
States
ASTRAP ENAMAP
2 0.7
17
25
24
5
21
31
52
36
5
7
13
0
19
27
71
35-
RCDM MCARLO
4 0.8
8
15
10
7
9
1-3
21
12
3
5
5
1
5
6
7
6
(-
**
*Background of 2 Kg.S.ha.~1yr.~ added to MOE values
(1) Source: Interpolated from a map prepared by the National Atmospheric Deposition
Program (1981) based on data for March 1979 to March 1980.
-------
7-15
models to overestimate deposition near large emission regions.
The models show an average annual wet sulfur deposition rate of 13
kg.S.ha.~%r."^ at the Adirondacks sensitive area, which
compares very well to the observations. The standard deviation of
the model results is 7.2 kg.S.ha.~%r."^ which reflects the
wide range in model values. Table 7-6 shows the Phase I and II
estimates by the OME-LRT and AES-LRT models are the same while the
estimates by the ASTRAP, ENAMAP and RCDM models changed significant-
ly due to the use of 1978 meteorology and the Phase II S02 emissions
inventory and some changes to the input parameters. In general/
the Phase II-estimates by the OME-LRT, MEP-TRANS and MCARLO models
+
and within -50% of the observed values, while the estimates by the
+
AES-LRT and RCDM models are within -75% of the observed values.
The ASTRAP and ENAMAP model estimates are generally much higher
than the observed values at all the targeted sensitive areas. All
the models show a general tendency to over-predict the observed
wet sulfur depositions in the targeted sensitive areas except for
the MCARLO model. These preliminary evaluation results will be
used by the modelers to refine their input parameters and check
their results before starting the second round of model evaluation
(see Section 6).
-------
Chapter 8
ANALYSES OF TRANSFER MATRICES FOR SULFUR OXIDES
8.1 Introduction
Annual transfer matrices from the different LRTAP models
were displayed in Chapter 7. These matrices represent mean
source-receptor relationships calculated by the models for
various meteprogical periods and then normalized and averaged
to be expressed as annual matrices. When computing the
matrices/ the values of the model parameters were chosen
independently for each model. Also/ no attempt was made to
"tune" the models with measurements to obtain the "best fit"
parameters.
Before using transfer matrices in the assessment iterations/
it is necessary to determine the seasonal variability and
sensitivity of the transfer matrices to uncertainties in the
model parameters. In this chapter, some preliminary results
on seasonal variability obtained with the ASTRAP and the
AES-LRT models will be presented in the next section. In
section 8.3, sensitivity to variations to input parameters
are examined for the AES-LRT, OME-LRT and the MEP-TRANS models.
8.2 Seasonal Variations
8.2.1 AES-LRT Results
The annual transfer matrices from the AES-LRT model were
computed using meteorological data for all of 1978 and the
-------
8-2
parameter values in Table 8-1. The parameter values,
taken as annual averages, were held constant throughout the
year, while in reality the parameters vary spatially and
temporally.
To test the adequacy of simulating the 1978 annual
concentrations and depositions with average parameter values,
the quantities were recomputed using seasonally varying
parameters. The deposition velocities and transformation
rates were increased during summer season. During the winter,
the washout ratio for sulfur dioxide was increased, while the
values for deposition velocity of sulfur dioxide, washout
ratio for sulfate and transformation rate were decreased as
shown in Table 8-1. The average parameter values were, assumed
to apply during fall and spring. Emission rates were held
constant throughout the year.
The differences between concentrations and depositions/
computed with the temporally varying parameters and fixed
parameters were site dependent, but small (see Tables 8-2
and 8-3). Concentrations of sulfur dioxide and depositions
of wet and dry sulfur for the two computations were within
10 per cent of one another.
-------
8-3
Table 8-1. AES-LRT Model Parameter Values Used
in the Sensitivity Study
Parameter
Dry deposition
VDSO (cm s"1;
Dry depositior
velocity of
1
i velocity of
S02
S04
Base Case
0.5
0.1
. Winter
0.2
0.1
Summer
1.5
0.6
VDS04) (cm s
Washout ratio S02:WS02 (dimensionless) 3.0'IQ4 3.0*105 3.0*104
Washout ratio S04:WS04 (dimensionless) 8.5'105 8.5'104 8.5*105
Transformation rate: k (% hr"1) 1.0 0.3 1.5
Mixing Height: H(m) (1) (2) (3)
(1): climatological monthly mean values for the entire year
(2): climatological monthly mean values for the season indicated
-------
8-4
Wable 8-2. Seasonal Variations in the AES-LRT Model Transfer Matrices
of Absolute Values (1978)
WET DEPOSITION OF SULFUR (kqS.ha."1)
PERIOD
Spring
Summer
Fall
Winter
Composite
Year
^se Year .
Per Cent
Difference
. BNDW.
0.43
. , .0.30 .
0.22
0.31
1.26
1.50
16
ALG.
3.66
. 1.46 .
2.88.
1.76
9.76
10.36
6
. MUSK.
. . 4.46 .
, 2.87
5.46
3.69
16.48
17.6.0
6
. . I
QUE.
2.01
1.31
2.37
2.29
7.98
9.02
12
RECEPTOR
. S.NSC.
0.98
.0...53
, 1.89.
.2.20
. 5.60
5.59 .
. 0
VT-NH .
2.57. ,
. 1.87
. 4.18. .
4.24.
12.86
13.10
2
ADIR. .
. 2.97 .
.2.18. .
, 5.02,
., 4.90
, 15.07
15.74 .
.4 . .
PENN.
6.13
. 7.42
9.24 .
13.06
35.86
33.51
-7
SMOKIES
. 3.92
2.39 .. ,
. 3.4.9 . .
. 8.11
1.7.91
16.66
. -8 . .
Average S09 CONCENTRATION Qag/m3)
PERIOD
Spring
Summer
Fall. .
Winter ...
Composite
Year
fcase Year
Per Cent
Difference
BNDW..
.4.2
1.2
., 2.3 . ,
. . 3.8,
2.9
2.9
2
ALG.
12.7 .
3.6
.1.0.1 .
1.1.3 .
9.4
10.3 .
8
MUSK..
24.1 .
. 8.8
.28.1
. 3,1.3 ,
23.1
.24.9
. 7
I
. QUE..
8.3.
2.9
11.3
11.2
8.4
9.2
8 . .
RECEPTOR
S.NSC
7.4. . .
3.2
. 11.7
17.2
9.9 .
i.o.o . .
1
VT-NH
13.9 ,,
7.4.
.20.4.
, 23.9
. 16.4
. 17.3 . .
5 ...
.ADIR. .
.16.3 ..
. . . 6.0
22.8 ,
28.2 . ,
18.3 .
. 19.3 .
, . .5
PENN..
47.1. , . .
28.6 .
.70.5
79.8 ,
.56.5
60.3
6
SMOKIES
. 30.2 ...
10.1 . .
.32.1 ...
. S2..4
3.1.2 .
31.2
0
-------
8-5
8-2. (continued)
AVERAGE S0 CONCENTRATION
PERIOD
Spring .
Summer ....
Fall, .
Winter .
Composite
Year
Base Year
^er Cent
Lfference
, BNDW.
1.9. .
0.4 ..
0.6
0.6
0.9 .
1.1
19.
. ALG.
.".'4.7
1.8
..2.8
1.7
2.8
3.5 .
.21 ,
MUSK. .
10..2 . .'
..4.7';
. 7.9
...3.9
6.7
8.4
.21
I
QUE.
. 3.6
.1.6
4.1 . .
2.2
.2.9
3.7
23
RECEPTOR
S.NSC
.3.6.
. 2.1
...5.9
4.0.
3.9.
, .5.2 .
25
VT-NH
. . 5.3
.. 3.0.
.7.2.
. . 4.0 .
. 4.9.
..6.4
23'
ADIR..
....6.8
3.8
9.9
. , . .5.2 ..
6.4 ..
. . . 8.5 . .
24
. PENN., .
12.3 .
. .8.8
1.6.8 . .
.9.0
11.7
15.3, , .
, 23 ,
SMOKIES
.13.6 . . ,
5.2
.. 1.3.7 . ,
. . 8.4.
. 10.2
.. 13.3 .
23 ,
TOTAL DRY DEPOSITION OF. SULFUR (kg.S.ha"!)
PERIOD.
Spring
Summer
Fall .
Winter
Composite
Year
Base Year
Per Cent
Difference
BNDW.. .
L. 0.88 .
0.77
0.46
. 0.32
2.43
. 2.44
0
ALG..
2.63
2.39
2.07
0.93
8.02
8.45
5
. MUSK.. .
, 5.02.
. 5.96 .
. 5.75
.2.57
. 19.3.0. .
20.49
. 6
I
.QUE..
1.73
. 1.96
2.34
, .0.94 ,
6.97
. 7.65
9
RECEPTOR
, S.SNC.
1.56
.2.2,5 .
..2.47
1.46
7.7,4 .
. 8.41 .
. 8 .
. .VT-NH .
2.88
. . 4.87
4.20
1.99
.. 13.94 .
.14.3.0 .
. . 3 .
. .ADIR.
3.40
4.13.
4.75
, . 2.36.
.1,4.64
. 16.13
. . 9 ..
PENN..
. 9.61
18.33
14.34
6.53. .
48.81
49.11
.1
SMOKIES
6.31
. . 6.80
.,6.6.8. ..
.4.35 .
24.14 .
.26.02
. . 7 , . .
-------
8-6
8-3. Seasonal Variation in AES-LRT Model Transfer Matrices of Per Cent of
Total or of Annual Average (1978)
WET DEPOSITION OF'SULFUR AS PER CENT OF ANNUAL
PERIOD
Spring. .
Summer
Fall . .
Winter.
. BNDW , .
... 34
24
L 17
. . 25 ,
. ALG..
.38
15
30
18 . .
MUSK.
... .2.7 . .
17
33
. 22 .
I
. .QUE.
. . .25 . .
16
30
29
RECEPTOR
. S..NSC
. 18 . .
9
34
39
VT-NH
. , 20
15
, 33
33
ADIR..
20
14
33
. 33
PENN..
. 17
21,
26
36
. SMOKIES
22
.. 13 .
, 19,
. , 45 ....
S02 CONCENTRATION AS PER CENT OF ANNUAL. AVERAGE
Spring^
Summer
Fall
Winter, .
.BNDW..
. 147
41
78
.133
ALG.
13.5-
,'38 .
108
.120 . . .
MUSK.
104
. 38 . ,
122
136
I
QUE.
98 .
. 35
. .1.3.4. . . .
133
IECEPTOR
S.NSC
75
33
, . .119, . . ,
174
VT-NH
, 8.5 , ,
45
124
146
ADIR.
89,,
33
124
154
PENN.
83, .
. ,51 .,
. 125 .
141
SMOKIES
, 97
. . . 3.2 , . . ,
103
168
SO4 CONCENTRATION AS PER CENT OF ANNUAL AVERAGE
PERIOD
Spring
Sunnier .
K.
.er .
BNDW.
211
46
71
70
ALG.
170
., 65
102
62
MUSK.
153
. . 70. .
118
58
I
QUE.
126
54
143
77
RECEPTOR
S.SNC
192
55
151
103
VT-NH
109
62
147
,82
ADIR.
106. ,
59 .
154
81
PENN.
105
75
143
77
SMOKIES
133
51
134
, ,82
-------
8-7
8.2.2 ASTRAP Results
The ASTRAP model has been run using trajectory statistics
derived from January and July 1978 meterology. The ASTRAP 1978
annual transfer matrices are based on an arithmetic mean of
the January and July matrices. Currently, no conclusion can be
made regarding the suitability of developing an annual average
concentration or total deposition from an average of just
two months.
The ASTRAP January and July matrices reflect not only
differences in the monthly meteorology, but also have been derived
using seasonally and diurnally varying emissions, emission release
heights, dry deposition velocities, conversion rates for S02
and 304 and eddy diffusivities. Table 8-4 summarizes the seasonal
variation of the ASTRAP.model parameters.
Table 8-5 presents the ASTRAP model simulations of monthly
average SC>2 concentrations and total wet sulfur depositions for
January and July 1978. This table shows there were significant
variations in ASTRAP simulations between these 1978 winter and
summer months. The ASTRAP monthly average SC>2 concentrations
for January at the nine receptors are on the average about 40%
higher than the July concentrations. On the othr hand, the
wet sulfur depositions do not show a consistent difference
between January and July. The higher S02 concentrations in
January can be attributed in part to the reduced dry deposition
velocities, and seasonal varying emissions amounts and distributions.
-------
8-8
Table 8-4.
Seasonal Variation in Parameter Values for the
ASTRAP Model
Parameter
V(jS02 average
(cm s~l)
V^S04 average
(cm s~l)
k average (% hr"1)
January
0.25
0.28
0.48
July
0.41
0.4.6
2.00
-------
"Cable 8-4. January and July 1978 ASTRAP Model Estimates of SC>2 Concentrations
and Total Wet Sulfur Depositions at the Nine Receptors
Average. SQ9 Concentration, ,(M9
_o
January
July
% Deviation
[Jan-July]
Jan
January
July
% Deviation
Boundary
Waters
0.4
0.2
50
Boundary
Waters
0.07
0.08
-14
Algoma
16.1
10.0
38
Algoma
0.90
0.60
33
Muskoka Quebec
13.3 7.0
8.7 6.7
35 4
.TOTAL WET SULFUR
Muskoka Quebec
1.25 0.71
1,08 1.37
14 -93
So.NS
2.8
1.3
54
.DEPOSITION.
So.NS
0.18
0.23
-28
Vt-NH Adirondacks
7.8 7.6
4.6 4.7
41 38
(kg.S.ha."1)
Vt-NH Adirondacks
0.88 1.66
0.99 1.61
-13 -28
W. Penn. Smokies
28.9 25.1
14.1 14.2
51 43
W. Penn. Smokies
1.93 1.52
2.55 1.55
-32 -2
00
-------
8-10
In addition/ the wet deposition is obviously closely correlated
with the pattern of precipitation amounts in the months under
study, and thus would not necessarily to show any correlation
«
between the months.
The sensitivity of ASTRAP to the mean value or diurnal
variation of its parameters has been assessed in terms of changes
in sulfur budget or changes in predicted concentration and
deposition maxima. Since these model products are not currently
under consideration, we direct the reader to the ASTRAP Model
Profile for details.
8.2.3 Summary
Although these preliminary results do not yet permit any
general conclusions to be drawn on the importance of seasonal
variations in concentrations and. depositions, the results indicate
that seasonal variability can be significant and should be inves-
tigated further.
-------
8-11
8.3 Model Sensitivity to Parameters
The sensitivity of modeled sulfur concentrations and depositions
to change in values of input parameters has been investigated.
»
Model parameters such as deposition velocities/ wet removal,
conversion rates of sulfur dioxide to sulfate and mixing heights
were varied, one at the time, within the range normally used for
long-range transport modeling.
Different methods were used by the different modeling groups.
The AES-LRT model used actual meteorology and emissions and
studied the change at the targeted sensitive receptors. The OME-
LRT and MEP-TRANS models used a single hypothetical source and
investigated the response at various distances from the source.
In the OME-LRT work the meteorology was simulated, while in the
MEP-TRANS work, the actual meteorology was used. The results
have been expressed either as per cent deviation from the base
case value or as sensitivity indices. The sensitivity index
gives the fractional change in computed concentration or deposition
(A) as function of factional change in parameter values (P),
defined as d(lnA)/ d(lnP). Highlights from the evaluations are
given below while details are provided in the Model Profiles
(see Appendix 5).
-------
8-12
8.3.1 AES-LRT RESULTS
The sensitivity of the modeled concentrations and deposition
amounts to changes in the values of input parameters were investiga-
«
ted at the sensitive receptors. The emissions inventory, meteorology
and parameter values were the same as those used to compute the
AES-LRT transfer matrices presented in Chapter 7.
Dry deposition of sulfur dioxide and wet deposition of sulfur
are the principle removal mechanisms for sulfur in the base case.
Thus, changes in dry deposition velocity of sulfur dioxide
2
the transformation rate of sulfur dioxide to sulfate (k) and the
washout ratio of sulfur (W) results in the largest change in
wet deposition of sulfur. This is clearly illustrated in Table
8-6 by the magnitude of the sensitivity indices in the rows
W.
2
As an example of the changes in concentrations and deposition
at an individual site, the results (see Model Profile for rest of
figures and tabulations) for iMuskoka are:
Sulfur dioxide concentration is most sensitive to
changes in dry deposition velocity of sulfur dioxide
(index: - 0.61) and mixing height (index: -0.25 and 0.54),
Sulfate concentration is most sensitive to changes in
dry deposition velocity of sulfur dioxide (index: - 0.40)
and the rate of transformation of sulfur dioxide to
sulfate (index: 0.88),
Dry deposition of. sulfur dioxide is most sensitive to
changes in dry deposition velocity of sulfur dioxide
(index: 0.39) and 'mixing height (index: -0.25 and -0.54),
-------
Table 8-6. ^Bisitivity Index - Fractional Change in Wet Depo^Pion As a Function of
fractional Change in Parameter Value - Annual (d In Dep/d In Parameter)
i-i-i-i-i-------------'----------------'-. '-<:-..:.:----i-:~--------$.:---:.:.:--i----i----:.'^.
Parameter BNDW ALG MUSK QUE SNSC VTNH ADIR PENN SMOKIES
and, flange. .... ,...'. . /,; .^/^'.'. /.^ .^\^V_/.V//^.^^ .'.'.'.'.',',
VdS02 -0.55 -0.52 -0.50 -0.59 -0.64 -0.53 -0.53 -0.40 -0.49
0.2-1.5 cm/s
VdS04 -0.14 -0.16 -0.15 -0.17 -0.28 -0.15 -0.16 -0.12 -0.19
0.1-0.6 cm/s
K 0.49 0.55 0.58 0.56 0.62 0.54 0.57 0.55 0.58
0.3-1.5%/hr
WS02 0.07 0.06 0.06 0.05 0.04 0.07 0.06 0.08 0.06
3 x 103 - 3 x 104
NA/SC-2 0-21 0.16 . 0.18 0.15 0.11 0.18 0.16 0.28 0.21
3 x 104 - 3 x 105
S04 0.34 0.39 0.41 0.37 0.46 0.36 0.39 0.42 .49
8.5 x 104-1.7 x 106
H (Mixing Height) 0.22 0.12 0.13 0.28 0.37 0.17 0.22 -0.11 -0.05
(0.5 x Base H)-
Base H
H (Mixing Height) 0.0 -0.14 -0.15 0.01 , 0.04 -0.05 -0.08 -0.31 -0.24
Base H-(2x Base H)
-------
8-14
Wet deposition of sulfur dioxide is most sensitive to
changes in dry deposition velocity of sulfur dioxide
(index: -0.70) and washout ratio sulfur dioxide (linearized
index: 0.93 and 0.67), t
Dry deposition of sulfate is most sensitive to changes in
dry deposition velocity of sulfur dioxide (index: -0.40'),
dry deposition velocity of sulfate (index: 0.08) and rate
of transformation of sulfur dioxide to sulfate (index:
0.88),
Wet deposition of sulfate is most sensitive to changes
in dry deposition velocity of sulfur dioxide (index:
-0.46), rate of transformation of sulfur dioxide to
*
sulfate (index: 0.86) and washout ratio of sulfate (0.57),
Dry deposition of sulfur is most sensitive to changes
in dry deposition velocity of sulfur dioxide (index:
0.35) and changes in mixing height (indices: -0.25 and -0.53),
Wet deposition of sulfur is most sensitive to changes in
dry deposition velocity of sulfur dioxide (index: -0.50),
rate of transformation of sulfur dioxide to sulfate
(index: 0.58) and washout ratio of sulfate (index: 0.41),
Total deposition of sulfur is most sensitive to changes
in mixing height (index: -0.34), the sulfate washout
ratio and transformation rate (indices: 0.15),
-------
. 8-15
8.3.2 QME-LRT. .R.e.su.ltS
The sensitivity of the OME-LRT model was investigated for
an ideal case rather than at the sensitive receptors. The idealized
source-receptor geometry, depicted in Figure 8-1, enables the testing
of the model sensitivity to the input parameters as a function of
the source-receptor orientation independent of actual meteorology.
The model parameters evaluated, apart from dry deposition velocity
of sulfur dioxide and sulfate, conversion rate of sulfur dioxide to
sulfate, mixing height and wet deposition rates, include meteorolo-
gical parameters such as the length of wet and dry periods, wind
speed and direction, and horizontal dispersion in the. along-wind
and cross-wind directions. The sensitivity indicies for wet deposi-
tion from a source at the different receptors are shown in Table 8-7.
In summary:
Dry conversion rate (k^):
For receptors down wind of the source the calculated wet
deposition rate is insensitive to values of ^(.6% to 1.2%)
except at large distances (/-*-1000 km). The result is the
same at cross wind receptors. Elsewhere the calculation
is insensitive to the value of k^.
Wet conversion rate (kw):
At near source receptor sites the deposition is sensitive
to values of k., greater than 5% hr'1. Below this conver-
W
sion rate there is no significant sensitivity. The con-
version rate must be large to show sensitivity due to the
-------
8-16
12 m
18
15 *x
\
\
\
\
\
\
IK
\
\
\
\
\
\ lOo
\
v
" e -^
17 16
9
/
/
f
f
/
/
s
s »
f m
8 ,-''' ,-''''' 6
/ ,'''
' ,''
7 / ,-"*'''
f O ^4
V 9 (3 .- «. ... ^
1 2 3
mean wind direction
source
Scale
\ .]
200 km
Figure 8-1. Idealized Source-Receptor Geometry Used for
OME-LRT Model Sensitivity Studies
-------
Table
OME-LRT Model Sensitivity of the Wet Sulfur ition Factor for the Idealized
Source-Receptor Geometry Shown in Figure 8-1. (The sensitivity is presented as
the ratio of the fractional change in the deposition factor with respect to the
fractional change in the parameter).
Wet Sulfur
Deposition
Factor
(10-3 gS/m2/yr)/
(ktonne S02/yr)
PARAMETER (base case value)
KD (1% hr-1)
KW (1% hr"1)
VD (1 cm/s)
VDBAR (0.1 cm/s)
LAMDA (3.0 x lO^s"1)
LAMDAB (1.0 x lO^s"1)
MIXHT (1000 m)
FD (0.9)
FW (0.1)
UM (10 m/s)
VM (6.0 m/s)
WIND (10 m/s)
WDIR (270°)
TAU D (46 hr)
TAU W (7 hr)
Ri
5.890
(x 10~13)
0.037
0.680
-0.180
-0.003
0.560
0.052
0.169
-3.100
0.300
-1.870
-9.020
0.452
-1.240
-0.546
0.316
.R2
1.710
(x 10 i3)
0.089
2.160
-0.343
-0.009
0.384
0.063
0.336
-0.770
0.090
-0.113
-0.895
0.551
-2.040
-0.724
0.386
K3
0.580
(x 10~13)
0.165
-1.670
0.530
0.020
0.257
0.075
0.540
0.170
-0.020
-3.050
-0.865
0.648
-3.430
-0.722
0.344
W-.2
0.051
(x 10~13)
0.330
-2.200
-0.780
-0.080
0.086
0.063
0.870
0.410
-0.040
-0.094
1.090
0.998
-3.480
-0.469
0.174
R15
0.044
i (x 10 13)
0.320
-2.130
-0.770
-0.070
0.091
0.064
0.860
0.400
-0.040
0.117
0.194
-1.440
0.454
0.478
0.180
JU.8
0.049
(x 10"1)
0.290
-1.880
-0.730
-0.060
0.119
0.069
0.810
-0.390
-0.040
3.040
-0.891
-1.780
0.213
-0.531
0.221
00
-------
. 8-18
magnitude of other pathways for wet deposition (e.g. wet
deposition of SC^). At large distances wet deposition
is not sensitive to the wet conversion rate - again other
pathways dominate.
Dry deposition velocity of SC>2:
Wet deposition is sensitive to this parameter at all
locations/ particularly at large distances. This is due
to the loss of sulfur by day deposition for subsequent
wet deposition.
Dry deposition velocity of 804:
For the range of 804 dry deposition velocities considered the
estimate of wet deposition does not depend on this parameter
(insignificant pathway).
Wet deposition rate of SC^:
Wet deposition is very sensitive to this parameter downwind
and near the source. This pathway near the source can
result in greater than 80% of the wet deposited sulfur to
be SC>2. Far from the source wet deposition is less
sensitive to this parameter as other pathways to wet deposi-
tion become important and SC>2 has been depleted or converted.
Wet deposition rate of 804:
This pathway for wet sulfur deposition is marginally sensi-
tive to the value of deposition rate as the total amount
of wet 804 available for scavenging is controlled by other
model parameters (e.g. kw
-------
8-19
Mixed layer height:
Due to the assumption of vertical homogeneity an increased
mixed layer height reduces the ground level concentration
thereby reducing the loss by dry deposition while the wet
scavenging rate remains constant. Therefore at all loca-
tions the wet deposition rate increases as the mixed height
increases.
Fraction of time (Eulerian) of dry (f^) or wet (fw) conditions:
The estimated wet deposition is sensitive to these para-
meters close to the source downwind. These locations
are near enough to the source that the nature of the
pollutant (wet or dry) emitted is not 'forgotten1. (i.e.
travel time is less than the time taken to lose the particles
identity ("Crw or tr^) For these regions, if we increase
the fraction emitted wet then wet deposition increases propor-
tionally.
Lagrangian duration of wet (T^) and dry (Tr^) periods:
The model estimate of wet deposition is sensitive to these
parameters for all source-receptor orientations as the
parameter control rate of exchange between wet and dry
particles.
Horizontal/ along wind dispersion ( 6~u):
The model estimate of wet deposition is sensitive to this
parameter as it controls the probability of the emission
reaching a receptor. Least sensitive are the directly down-
wind sites which do not rely on dispersion to transport
-------
8-20
the emissions. Conversely upwind sites depend solely on
dispersion and are the most dependent on this parameter.
Horizontal, cross-wind dispersion ( £~~v):
All receptor sites are sensitive to this parameter. As
for (TU' this parameter controls the value of the probability
of arrival at the site. Sites least affected are those downwind
near source sites in the cross-wind direction that are
due to depletion of the pollutants by cross-wind spread.
Mean wind speed:
All receptor sites are sensitive to this parameter. Wet
deposition at a site will increase or decrease with increased
wind speed depending upon whether increased wind speed will
move the pollutant toward or away from-the receptor more
quickly.
Wind direction:
All downwind sites are sensitive to the wind direction;
however, for the range chosen (255° - 285°) directly up-
wind sites do not show any sensitivity. Deposition is
maximum for those sites directly down wind.
Additional results are provided in the model profile.
8.3.3 M.E.P TRANS Results
The sensitivity analysis was carried out for a single source
by annual runs with the range of parameter values indicated in
Table 8-8, varying only one parameter at a time. The results for
-------
Table 8-8. Range of Parameter Variation for the MEP-TRANS Model Sensitivity Analysis
Parameter
Sulfur Nitrogen
Summer Winter Summer Winter
Low Std. High Low Std. High Low Std. High Low Std. High
Primary Deposition
Velocity (cm/s)
0.1 0.8 2.0 0.1 0.3 2.0 0.1 0.5 2.0 0.1 0.3 1.0
Secondary Deposition
Velocity (cm/s)
0.1 0.4 2.0 0.1 0.3 1.0 0.1 0.5 2.0 0.1 0.3 1.0
Transformation Rate
(% hr -1)
Primary Washout Rate
(hr -1)
1.0 2.0 4.0 0.5 1.0 2.0 3.0 5.0 10.0 1.5 2.0 4.0
Not Varied
0.01 0.04 0.1 0.01 0.04 0.1
T
|SJ
Secondary Washout Rate 0.04 0.5 0.8 0.04 0.5 0.8 0.04 0.5 0.8 0.04 0.5 0.8
(hr -1)
-------
8-22
the sensitivity of 804 wet deposition as an example of the analysis
performed for the parameters are presented in Table 8-9.
In Table 8-8, the average 864 deposition is the average of
all receptors within each of the distance ranges indicated. Thus,
the average deposition within 100 km from the source is 42 mg
S/m^/yr for the standard choice of the parameters. The percentage
change is then the average over the receptors within the given
distance range. Thus, choosing the low value for transformation
rate reduces the wet deposition in the near range by 47%, in
the mid-range the reduction is only 34% and in the far range it
is only 23%. The high transformation rate increases the washout in
the short-range by 80%, but has little effect in the long range.
This reflects the fact that removal of 804 by precipitation is more
efficient than SC>2 removal. Reducing the deposition velocities,
particularly the S02 deposition velocity, increases the 804 wet
deposition; in this case the effect being cumulative with distance.
The effect of 804 washout rate on 864 wet deposition would seem to
be contradictory in that a reversal of the expected effect is
observed. Lowering the washout rate does lower the wet deposition
in the short range, but increases it in the far range, since the low
washout rate allows accumulation of sulfate which can be washed out
at longer range.
-------
Table 8-9. Sensitivity of Wet 804 Deposition to Variations in MEP-TRANS Model Parameters
Distance
From Source
. , , .(km)
0 -
100 -
200 -
500 -
700 -
1000 -
>
100
200
500
700
1000
1500
1500
Average 804
Wet Dep. (mg S m~2)
. Standard, Parameter, .
42.
20.
6.
1.
0.
0.
0.
3
2
9
9
7
2
1
Transformation
Rate
.Low , fligh. . .
-47
-45
-39
-34
-30
-23
-23
+80
+72
+47
+30
+19
-2
0
Deposition Velocity
so2 so4
.... .Low High . , , .Low. . . , High
+18
+27
+55
+77
+98
+140
+143
-26
-32
-46
-54
-60
-72
-75
+5
+7
+13
+21
+20
+22
+21
-16
-20
-31
-43
-40
-40
-39
Washout Rate
of SO,
. . LOW. ttigh.
-52
-38
+28
+87
+173
+433
+407
+3
-2
-11
-21
-22
-21
-25
00
-------
8-24
This example points out the complex dependence on the several
removal and transformation processes within the model, and shows
that one cannot consider the various parameters as being truly
independent.
For further results see the Model Profil-e.
8.3.4 Other Model Results
Sensitivity analyses for the RCDM, ENAMAP, CAPITA Monte Carlo
and UMACID models will be provided in an addendum to the Modeling
Subgroup Report by September 1981.
-------
Chapter 9
PRELIMINARY SOURCE-RECEPTOR
RELATIONSHIPS FOR NITROGEN OXIDES
9.1 Introduction
At the present time the basic principles of the atmos-
pheric chemistry of oxides of nitrogen/ at least with respect
to the conversion from primary pollutant (NO) to the ultimate
secondary pollutant (HN03) are reasonably well characterized
via experimental smog chamber studies. The evaluation and
verification of these processes under atmospheric conditions
are far less established. The role of ozone and peroxy-radicals
in converting NO to N02 is well established/ and the subsequent
formation of HN03 is known to be due to the reaction between
OH radicals and N02. It is also well documented that/ con-
currently with HN03 formation, PAN production will occur pro-
vided that sufficient acetaldehyde is available. Although
actual observations concerning PAN in the atmosphere are
limited in number, they do suggest that under most conditions
when NO is emitted from a source, such as transportation
activities, enough hydrocarbons are indeed co-emitted to
cause PAN formation. Finally, from our present understanding
of these processes it appears that most of the reactions occur
exclusively in the gas phase and hence/ contrary to the SOX
chemistry, one is not burdened by the well-known complications
caused by liquid phase chemistry in the atmosphere.
-------
9-2
It is at present difficult to model the NOX chemistry
with the same confidence as the SOX chemistry for the
following two reasons:
Firstly, the SOX chemistry is simplified to the ultimate
limit with the linear parameterized oxidation reaction S02>
sulphate. From a chemical viewpoint it can be argued that
this is feasable since the conversion is relatively slow
(S02 has a chemical lifetime of a few days) and hence
averaging of diurnal variations is acceptable.
The common use of a timestep of more than one hour in
model runs is then also reasonable. On the other hand, the
NOX oxidation chemistry takes place at faster rates, (the
oxidation of NO» N02 within the timespan of a few hours,
while the chemical lifetime of N02 is less than one day)
and consequently the use of diurnally averaged rates as well
as the application of timesteps of a few hours is questionable.
In connection with this matter, it can be stated that
the SOX oxidation process has a limited influence on the
concentration of oxidizing species (hence first order chemistry
can be assumed), while the NOX oxidation process has a
strong feedback effect on the levels of oxidizing species.
-------
9-3
The second major problem rests with the evaluation
of the modeling efforts. The simplified SOX chemistry in
the current LRTAP models can at least be defended with the
qualified statement that comparison with data from actual
observations in the field shows that the modelled data are
not very divergent.
Such an evaluation of potential NOX chemistry modeling
is not possible at the present time since the data base on
field observations is very poor indeed. The networks that
have been in operation to date have produced data on wet
deposition of nitrate, but it is highly doubtful whether
they represent actual levels in the rain due to factors such
as liquid/gas exchange of the nitrate ion, and in some cases
biological activity in the samples. They produce no data on
levels of particulate nitrate or gaseous HNC>3 in the U.S.
The data from the APN and Ontario Hydro networks have unknown
(but potentially very large) errors. Finally, nowhere do
the networks produce data on concurrent levels of NOX or
PAN. Hence it is not even.possible to start evaluating
whether the complications that are anticipated from our
knowledge of the NC>2 chemistry are indeed serious enough to
make NOX modeling with the current LRTAP models impossible
or not. In addition, N02 measurements (when available at
-------
9-4
rural monitoring sites) are typically below the lowest detect-
able limits of currently available automated monitoring
equipment.
There is still another matter which causes NOX modeling
to be almost pure speculation at the present time. As can
be grasped from the review by Lusis and Shenfeld (1981) on
the seasonal effects on chemistry parameterization, there
is at least a limited amount of information available to
make guestimates on how to treat the deposition processes
of S02 and sulphate in the models. Similar information is
virtually nonexistant for the nitrogen oxide species, and
one is forced even more to use chemical and physical intuition
when parameterizing loss processes from the atmosphere.
^ 2 .P.a.r.ain.e.t e r i z a.t,i.on
Three preliminary efforts to model the chemistry of nitro-
gen oxides with LRTAP models appear to have been made. A
"minimal" chemistry adequate according to Jeffries (1979),
based on smog chamber studies, is shown in Figure 9-1.
Obviously LRTAP models cannot now incorparate such chemical
detail. The CAPITA Monte Carlo simulation model (Patterson
et al. 1981) has been so far the most extensive in its
chemical considerations by applying a scheme shown in Figure
9-2. The CAPITA model relies upon diagnostic determination
-------
Figure 9-1. Gas Phase Nitrogen Oxides Chemistry
(Jefferies, 1979).
(RQNQ)
. <-£
mo NO
RONQj
U1
-------
9-6
HN03
PAN -
HN03 DrY Deposition
HNC>3 Wet Deposition
^»-PAN Deposition
N02 Dry Deposition
'NC>2 Wet Deposition
Figure 9-2. The Kinetic Scheme for NOX Modeling Used By
the CAPITA Model.
-------
9-7
of rate parameters using ambient concentration and source
inventory data. Since ambient data are virtually nonexistent,
the applied rates were selected with consideration of the
overall chemistry and numerous other facts and hints that are
not documented (see Table 9-1)/ but very little information
was provided to support the quantitative values used. This
in mind, the following results were obtained from the different
models.
The MEP-TRANS model further simplifies the chemistry
by using the the first order reaction N02-WJ03 as the only
chemical reaction. Rate constants and deposition parameters
are presented without discussion on the choice of parameters
(see Table 9-2). The AES model also simplifies the chemistry
to the single reaction NC>2>NC>3 (Bottenheim, 1981).
Here it is argued that NO»NC>2 conversion is too
fast for the timeframe of the model to warrant separate
consideration, while PAN chemistry is deleted for lack of
any field data. Conversion rate is derived from the theoretical
rate constant ratio kHO+NO AHO+SO / assuming the majority
of SC>2 oxidation to be due to gas phase chemistry which
results in a conversion rate of 10%h~^. Deposition parameters
are from the sparse literature, see Table 9-3. For source
emissions all three models have used the same inventories,
i.e. those reported by Voldner et al., (1980) for Canada,
and Clark (1980) for the eastern U.S., which are presumably
good for 1978.
-------
9-8
Table 9-1. NO Kinetic Rate Constants (h"1) Used By
the CAPITA Model.
TRANSFORMATION
DEPOSITION
DN02 DN03 DPAN
July
October
January
April
Fast
Slow
Fast
Slow
Fast
Slow
Fast
Slow
0.
0.
0.
6.
0.
0.
0.
0.
5
1
4
08
25
05
4
08
0.
0.
0.
0.
0.
0.
0.
0.
04
02
03
015
015
008
03
015
0.
0.
0.
0.
0.
0.
0.
0.
04
02
03
006
02
004
03
006
0.
0.
0.
0.
0.
0.
0.
0.
02
01
015
008
01
005
015
008
0.06
0.03
0.05
0.025
0.03
0.05
0.05
0.025
0.02
0.02
0.015
0.015
0.01
0.01
0.015
0.015
Fast 0.038 0.029
YEAR AVERAGE
Slow 0.78 0.015
0.03
0.015 0.048 0.015
0.009 0.008 0.024 0.015
-------
9-9
Table 9-2. Parameter Choice For the 1978 MEP-TRANS Model
Simulations of NOX.
Parameter Nitrogen
Summer Winter
Primary Deposition 0.5 0.3
Velocity (cms"-'-)
Secondary Deposition
Velocity (cms"1) 0.1 0.1
Transformation
Rate (% hr"1) 5.0 2.0
Primary Washout
Rate (hr"1) 0.04 0.04
Secondary Washout
Rate (hr"1) 0.3 0.3
Mixing Height (m) 750 500
Table 9-3. Deposition Parameters For NOX Chemistry
Used in the AES-LRT Model.
'
cms"1 W
N02 0.2a 1/4 x WSQ b
HN03 1.0C 1/2 x WHS0
a Average from Bottger et al., 1978
b Beilke, 1970
c Estimate based on no surface resistance
d Estimated from work of Levine and Schwartz (1981)
W Scavenging ratio (dimensionless)
-------
9-1-0
9.3 Mode.l.ing/ re.suIts
It will be clear by now that with so little data avail-
able to anchor the parameterization, any modeling runs to
date cannot be considered to yield more than some educated
speculation of the role of nitrogen oxides in LRTAP. With
CAPITA, mp.de.1
The CAPITA model has not been run for specific source-
receptor combinations, but overall budgeting has been
attempted for the total Eastern North American continent.
It is predicted that on a yearly average basis 50-65% of the
total emitted nitrogen is transported out of the region con-
sidered (i.e. Eastern U.S. and Canada combined), the largest
export occurring during the winter (least HNC>3 formation).
With the kinetics employed, 17-20% of the emission is exported
as HNC>3, 10-16% in the U.S. and 4% in the Canadian region
considered. As far as transboundary flow is concerned, the
model predicts that of the total Eastern North American
emissions, 11-12% is transported from the U.S. to Canada, of
which about 8% is deposited (the remaining 3-4% is transported
further beyond the region considered). In contrast, 2% is
exported from Canada to the U.S. of which 1% is deposited.
Similar figures are predicted for net export of nitrogen and
HNC>3 deposition. In summary, then, transboundary flow of
nitrogen from the U.S. to Canada exceeds the Canadian nitrogen
export by a wide margin, according to the model.
-------
9-11
HEP-TRANS model
The MEP model has been run to produce annual average
concentration and deposition fields for 1978. High concen-
trations for NC>2 are predicted for the Detroit and New York
areas/ while N03 (as HNC>3 or as a component of TSP) peaks
in the New York area only. Total N deposition is predicted
in excess of 1.5 g m~2y-l in the lower Great Lakes region.
The MEP model has also been used to produce nitrogen transfer
matrices. The transfer matrix for total N is presented in
Table 9-4.
It is interesting to note that even if data on nitrogen
oxides are fragmentary and uncertain at this time that with
the simple nitrogen chemistry used, the wet NC>3 to 804
deposition in Southern Ontario and Quebec are in a ratio of
1 to 2, in agreement with a number of experimental studies
carried out in these areas.
the'AES-model
The AES model has been run only with the total emission
inventory/ to produce monthly and yearly average airborne and
deposition data for 1978 at some of the sensitive receptor
areas. These results are presented in Figures 9-3 to 9-5.
The correlation between observed and predicted data/ Table
9-5, is generally poor/ although it is remarkable that the
best correlation is obtained for the two receptor areas
(Pennsylvania and Muskoka) which are closest to large source
areas.
-------
Table 9-4. Normalized Transfer l^Bix For Total Nitrogen From the MEP-TRANS
Model (kg.N.ha."1 yr.^- per Tg.N.yr."1) ..
_...._........,.,..., ...^..'. '..' V ' '..'.' ' '..' '.'..'.' .RECEPTORS;. '_'_/ '.'.'.'.'.'.' ' '_.'..'.'.' '//.'..'/./_.'_,'_.'.,'. '_,V._.'.
#1 #2 #3 #4 #5 #6 #7 #8 #9 #10 #11
REGIONS Q:NO2 , 3..^ Wft.tgrs. , Alg;.. . MusK ., ^ Jjuie .^ ^ _. JS. .JSUJSc.., Jfl1 . NJH .,^ Adirj.^ _gejryv gjngKiea FJ.Qr.idA
(x TG N/YR)
# 1 Michigan 0.469 | 0.15 | 2.37 | 2.38 | 0.35 | 0.16 | 0.47 | 0.63 | 1.44 | 0.27 | 0.03 | 0.07
# 2 111. ,Ind. 0.724 | 0.09 | 0.64 | 0.58 | 0.13 | 0.04 | 0.18 | 0.23 | 0.39 | 0.98 | 0.17 | 0.57
# 3 Ohio 0.606 | 0.02 | 0.25 | 1.97 j 0.32 | 0.16 | 0.48 | 0.74 | 3.96 | 0.70 | 0.06 | 0.04
# 4 Penn 0.303 | 0.01 | 0.12 | 1.11 | 0.66 | 0.46 | 1.40 | 2.33 |14.83 | 0.17 | 0.04 | 0.00
# 5 NY to Maine 0.906 | 0.01 0.06 | 0.63 | 1.63 | 1.68 4.16 | 4.57 1.02 | 0.02 0.03 | 0.00
# 6 Kent^Tenn 0.336 | 0.02 | 0.08 | 0.13 | 0.03 | 0.03 | 0.06 | 0.09 | 0.35 | 11.72 | 0.82 | 0.17
# 7 N.Va to NC 0.541 | 0.01 | 0.04 | 0.21 | 0.10 | 0.09 | 0.24 | 0.34 | 1.93 | 3.56 | 0.30 | 0.01
# 8 Rest of 1.189 | 1.79 | 0.67 | 0.25 | 0.05 | 0.03 | 0.06 | 0.09 | 0.13 | 0.50 | 1.04 | 1.05
East, us ..._._,......_.... _.^:^_._^^:__,_,^^^_,_.:\_^\,^^^_,_. .^^_._,, ^_.^._._^_l_.',_^^_.,^^_.^, ._,'.\.:,,_
# 9 Ontario 0.150 | 0.06 | 2.30 |11.76 | 1.44 | 0.44 | 1.93 | 3.55 | 3.28 | 0.16 | 0.01 | 0.01
#10 Quebec 0.131 | 0.02 | 0.20 | 1.50 | 9.61 | 1.62 |10.97 |18.55 | 0.28 | 0.02 | 0.00 | 0.00
#11 Atlantic 0.055 | 0.00 | 0.02 | 0.10 | 0.56 | 4.69 | 0.31 | 0.16 | 0.10 | 0.00 | 0.00 | 0.00
Provinces .^^^^^^^^^^^^^^_._^^^_.^^^^^^^_!_!_.^^^_._..^_,_._._.^^^^^^^^^^_,^^^^^_l_!_l_l_'.^_._._'.^_
#12 Sudbury 0.003 | 0.06 | 3.80 | 7.06 | 1.57 | 0.27 | 1.75 | 3.50 | 1.33 | 0.07 | 0.00 | 0.01
#13 West 0.061 | 3.61 | 0.39 | 0.08 | 0.03 | 0.00 | 0.03 | 0.04 | 0.01 | 0.04 | 0.01 | 0.08
Canada ., , '.'...'.'...'. '.'.'._._,_._.^.._._.....^^_«.- .^^^^-^-^-.' > -^-
-------
9-13
Table 9-5. Correlations Between Observed and
Predicted N03 from the AES-LRT Model,
(ATIKOKAN, CANSAP): r = 1.5 + 3.2
(WAWA, CANSAP): r = 2.3 - .27
(PETERBOROUGH, CANSAP): r = .89 + .91
(PETERBOROUGH, HYDRO): r= -.11 +1.07
(QUEBEC, CANSAP): r = 7.8 - 1.08
(SHELBOURNE, CANSAP): r = 1.26 - .18
(REST OF EASTERN U.S., MAP3S): r = 1.20 + .07
(PENNSYLVANIA, MAP3S): r = .11 + .36
(BOUNDARY WATERS, CALCULATED): r= .09 (n = 11)
(ALGOMA, CALCULATED): r = -.14 (n = 8)
(MUSKOKA, CALCULATED): r = .43 (n = 10)
(MUSKOKA, CALCULATED): r = .48 (n = 12)
(QUEBEC, CALCULATED): r = -.35 (n = 8)
(NOVA SCOTIA, CALCULATED): r = -.19 (n = 11)
(REST OF EASTERN U.S., CALCULATED): r = .25 (n = 11)
(PENNSYLVANIA, CALCULATED): r = .68 (n = 12)
-------
9-14
9.4 C.o.n.cius ipjh4 .a,nd. R.e.c.orhpi.e.hd.ait.i.on.s
Modeling of nitrogen oxides chemistry is in its infancy.
In fact the data discussed in the last section must be viewed
more as speculative exercises which require extensive further
research before resemblance with the real world can be assumed.
As stated previously: (1) the time and space scales of present
LRT models are not sufficiently resolved to adequately treat
the complex non-linear transformations associated with nitrogen
oxide chemistry, and (2) the alternate evaluation method
which is heavily used for the SOX chemistry, and which consists
of comparison with observational data, is impossible for NOX
chemistry due to the lack of data. The best statement that
can be made at the present time is that our scientific knowledge
is inadequate to quantify the NOX-HN03 problem. Regional
scale models are urgently needed to evaluate the time-space
scale for the nitrogen oxides chemistry. More reliable
observational data are especially needed. For policy considera-
tions, no quantitative recommendations from the LRTAP models
can be extracted at the present time. In a qualitative
sense it appears NOX-HNC>3 transport should be more limited in
average distance travelled than SOX transport, and hence
should be more of a sub-regional problem.
With respect to transboundary flow one can point to the
large source areas and prevailing atmospheric flows, and
-------
9-15
suspect that sources in the lower Great Lakes area will have
considerably more impact on Canada, but that the other
large source area (The Washington-Boston corridor) will
probably have a lesser impact because of the greater distance
from Atlantic Canada. Conversely the impact of Canadian
sources on the U.S. should be relatively small in view of
their smaller size, and should mainly come from the southern
Ontario region.
-------
Chapter 10
PRELIMINARY/S.O.U.RCE-RE.CE.PTOR .RELATI.ONSH.IPS
FOR PRIMARY SULFATES
10.1 ,In,t.r.od,uc.t.i.o.n
In the Phase I Interim Report, Work Group 3A recommended
that Work Group 2 consider assessing the relative contribution
to acid deposition of primary sulfate emissions from oil-fired
and coal-fired combustion sources compared with that from
secondarily formed sulfates. The Work Group decided to
first review the previous primary sulfate modeling efforts
as to their adequacy for use in Phase II. The two principal
primary sulfate modeling efforts that were reviewed were
those by Dr. J. Shannon of Argonne National Laboratory for
the Environmental Sciences Research Laboratory of the U.S.
Environmental Protection Agency and by PEDCO, Inc. for the
Morgantown Energy Technology Center of the U.S. Department
of Energy. In addition, Work Group 2, in cooperation with
Work Group 3B, reviewed the three principal primary sulfate
emission inventories for the eastern U.S., namely, the EPRI
(SURE Phase II), the PEDCO, and the EPA (Homolya). Work
Groups 2 and 3B concluded there were significant differences
between the three principal primary sulfate inventories for
the eastern U.S. which would have to be reconciled in Phase III
A primary sulfate emission inventory for Canadian sources
would also be prepared in Phase III.
The Work Group found some major deficiencies in the PEDCO
modeling (Szabo, et al.,1981) of primary sulfate emissions
-------
10-2
and, therefore decided to use the Shannon-Argonne National
Laboratory results in the Phase II report (Shannon, 1981).
Additional modeling work using the participating models
and improved data bases especially for primary sulfate
emissions is planned in Phase III.
Some of the major deficiencies in the PEDCO modeling of
primary sulfate emissions that the Work Group found were:
(1) the use of an incomplete SC>2 emissions inventory
only the top 50 SC>2 emitters representing less than
50% of the total eastern U.S. emissions were used
for the distant sources and SC>2 emissions for New
York, eastern Pennsylvania, New Jersey, and New
England were used for local sources;
(2) the use of a very high sulfur dioxide to sulfate
conversion rate due to catalytic oxidation processes
that may only apply to some sources at night;
(3) the use of a very simplistic box model which is not
appropriate for determining either short range or
long range impacts; and
(4) the use of modeling results which even if accepted
as correct, do not provide conclusive evidence that
local sources are the major contributor to acidity
in sensitive areas like the Adirondacks.
-------
10-3
10.2 Em.i.s.s.ioAS
Previous regional modeling efforts have assumed primary
sulfate emission factors typically of about 2%; i.e., 2% of
the pollutant sulfur emitted in the form of sulfate and the
other 98% in the form of SC>2. Since the average rate of
chemical transformation of 862 to sulfate in the free
atmosphere has been assumed to be about 1% hr~l, shortly
after emission regional model simulations show that secondary
sulfate from atmospheric transformation dominates primary
sulfate (e.g., Shannon, 1981).
However, some recent field studies involving measurements
of primary sulfate and S02 emissions from large package
boilers for apartment complexes or hospitals in the New York
City area during winter have indicated a primary sulfate
emission factor of 13.4% (Homolya and Lambert, 1980). The
boilers burned residual fuel oil of 0.3% sulfur content.
While the total sulfur oxide emissions from such low-sulfur
fuel sources are only a fraction of total sulfur oxide emissions
in the eastern U.S., the significantly larger primary sulfate
emission factor calls for a more detailed investigation of
the relative importance of primary and secondary sulfate in
areas with those package boilers.
The sulfur oxide emission inventory, compiled as part of the
Sulfur Regional Experiment (SURE) study, is classified by fuel
type (coal, residual oil, distillate oil, and miscellaneous)
for point sources larger than 10,000 tons SOX per years (source)
-------
10-4
or 1000 tons SOX per year (stack), and by source type
(residential, commercial, industrial, transportation, and
small point sources) for area sources on a grid with spatial
resolution 80 km. This makes the inventory quite convenient
for an examination of the relative contributions of primary
and secondary sulfate.
In Shannon's (1981) analysis the SURE emission inventory
was classified into five subsets for both winter and summer.
Miscellaneous large point sources were combined with large
point sources burning distillate oil, and the various area
source categories were combined, but split geographically
into northeastern and southwestern regions. The sulfate
emission factors were assumed to be independent of season,
but the emission totals, particularly for the area sources,
varied considerably by season, and thus separate model simula-
tions were run for summer and winter. Table 10-1 lists the
assumed emission factors and the emission rate for each
subset of sources. Here, the northeastern region is defined
as being north of 36 degrees latitude and east of 78 degrees
longitude. Weighting of the primary sulfate emission factors
by the emission total of each category produces an average
primary sulfate emission factor of 3.0% in summer and 3.6%
in winter. The seasonal change in the average emission
factor is due mostly to the large increase in northeastern
area sources which have the highest sulfate emission factor.
-------
10-5
Table 10-1: Sulfate Emission Factors and Sulfur Oxide
Emission Rates.-
SOURCE CATEGORY EMISSION RATE SULFATE EMISSION
(equivalent tons S02/day) FACTOR
summer winter
Coal point sources 42,000 40,000 1.5%
Residual oil point sources 4,t)00 4,000 7.0%
Distillate oil point sources 6,800 6,800 3.0%
Northeastern area sources 4,500 4,500 13.4%
Southwestern area sources . .19.,.3.0,0 .27,,.3.0.0 3.0%
76,600 88,400
10.3 Aff,TRAP Model. Re.s.ul.ts
Previous simulations with the Advanced Statistical Trajec-
tory Regional Air Pollution (ASTRAP) model (Shannon, 1980)
had attempted to account for some of the variation in primary
sulfate emission factors by adjusting the factor by emission
level (layer). The adjustment was not the same in all
simulations, but typical sulfate emission factors ranged from
3.0% in the lowest layer to 1.8% in layers above 400 m or so.
The weighted average factor was thus a bit more than 2%,
still consistent with the typical sulfate emission factors
in regional models mentioned above. In order to investigate
the principal sources of anthropogenic sulfate in greater
detail, the ASTRAP model was modified to calculate the norma-
lized vertical profiles of both primary and secondary sulfate,
-------
10-6
rather than a single normalized vertical profile of total
sulfate as in previous work. Primary sulfate factors were
included as a parameterization in the concentration and
deposition subprogram of ASTRAP, which was exercised
separately for each of the emission subsets in Table 10-1.
The same January and July 1985 meteorological data were used
as in previous ASTRAP applications.
The change in ASTRAP formulations described above
produce only minor changes in the simulations of average S02
concentration and of dry and wet deposition of total sulfur,
since the wet removal parameterization in ASTRAP is indepen-
dent of sulfur species, the dry removal parameterization is
quantitatively similar for each species, and because the
large relative increase (50-75%) in the average sulfate
emission factor leads to only a 1-2% relative reduction in
the S02 emission factor. Therefore, our detailed examination
focuses on sulfate concentrations.
ASTRAP simulations of the average surface concentrations
of primary and secondary sulfate resulting from all eastern
North America anthropogenic sources in summer and in winter
are shown in Figures 10-1 and 10-2 and 10-3 and 10-4, respec-
tively. Viewed from the broad perspective of the eastern
half of North America, secondary sulfate still dominates
primary sulfate. However, the. winter sulfate patterns in
New England and the eastern half of the Middle Atlantic states
-------
10-7
Figure 10-1,
Isopleths of Summer (July - August) Primary
Sulfate Concentrations (ug m~3) Simulated
by the ASTRAP Model (Maximum Concentration
3.7 ug m~3).
-------
10-8
V \
Figure 10-2.
Isopleths of Summer (July - August) Secondary
Sulfate Concentrations (ug m~3 ) Simulated
the ASTRAP Model (Maximum Concentration
14.9 ug m~3).
-------
10-9
Figure 10-3.
Isopleths of V^inter (January - February) Primary
Sulfate Concentrations (ug m~3) Simulated by the
ASTRAP Model (Maximum Concentration 7.7 ug m~3).
-------
10-10
Figure 10-4.
Isopleths of Winter (January - February) Secondary
Sulfate Concentrations (ug m~3) Simulated by the
ASTRAP Model (Maximum Concentration 6.3 ug m~3).
-------
10-11
show that primary sulfate concentrations are as large as secondary
sulfate concentrations in the Washington-Boston urban region.
In fact, in the New York City area, primary sulfate concentra-
tions are larger than secondary sulfate during winter. If
the area source resolution were improved below the 80 km
squares of the SURE emission inventory, gradients of sulfate
concentration between urban and rural areas in the Northeast
produced by superimposing the broad regional patterns of
secondary sulfate and the relatively spiked urban patterns
of primary sulfate would probably be even steeper.
Some additional insight to the sulfate patterns may be
gained from examination of separate simulations of point and
area sources. The large point sources, upon which most
regulatory emphasis rests, are generally elevated, some well
above the heights reached by nocturnal inversions, and are
concentrated in the Ohio River Valley. Because of the eleva-
tions of the point sources, the plumes are decoupled from
surface removal processes during most nights so secondary
sulfate can increase before the sulfate plume contacts the
surface the next day. In addition, the average sulfate
emission factor for point sources is quite low, because of
the preponderance of coal combustion. For these reasons,
one would expect the dominance of secondary over primary
sulfate from point sources. Indeed, the model results for
the summer and winter cases, respectively, shows that primary
-------
10-12
sulfate from large point sources is only a small fraction as
great as secondary sulfate/ on a regional basis. The disparity
is greater in summer than in winter, because of the higher
rate of S02 conversion during warm, humid weather.
When one examines the simulations of emissions from area
sources, however, a different pattern appears. First, the
area sources actually consist of a very large number of small
point sources which are normally emitted near the surface.
The ASTRAP model simulations assume that area emissions are
initially evenly dispersed through the bottom layer (0-100
m). Thus, the decoupling mechanism is unimportant and the
effect of primary sulfate is felt at the surface immediately.
Second, the area sources are more important during the winter
season, when space heating is required, and that is the
season when atmospheric transformation of SC>2 is least
effective. Examination of the sulfate concentrations from
area sources for summer and winter shows that in the northeastern
urbanized corridor, primary sulfate concentrations are about
half as large as secondary sulfate concentrations during the
summer, but more than twice as large as secondary sulfate
during the winter.
Mention should be made here that the ASTRAP model simulates
only anthropogenic sulfate from the area east of the Mississippi
River. The flux of anthropogenic sulfate from the west is
not included in ASTRAP so the gradient of sulfate in the west
-------
10-13
part is steeper than might be expected in nature. In addition,
there was no attempt to add a background sulfate concentration
for biogenic sulfate. In addition, the SURE inventory is also
known to underestimate eastern Canadian S02 emissions and
caution is required in the interpretation of simulation results
in areas impacted by eastern Canadian sources. A final and
most important caveat is that the accuracy of the above
analysis is dependent upon the accuracy of the assumptions
about sulfate emission factors in Table 10-1.
Source/receptor transfer matrices for primary and
secondary sulfate concentrations are shown in Tables 10-1 and
10-2. It should be noted that the emission data (Table 6-1)
and meteorological fields used in generating the transfer matrices
are not identical with the input data for the simulations
shown previously in Figures 10-1 through 10-4.
-------
10-14
Table 10-2. Transfer Matrix for January 1978 Sulfur Dioxide and Sulfate
Concentrations (ug/m3) from the ASTRAP Model Using the Phase III
State/Provinces SC>2 Emission Inventory and Primary Sulfate Emission
Factors in Table 10-1.
S02
Spurce Regions
1
2
3
4
5
6
7
8
9
10
11
Total
Primary 864
Source Regions
1
2
3
4
5
6
7
8
9
10
11
Total
1
0.01
0.03
0.01
o.oo
0.00
0.00
0.00
0.11
0.36
0.01
0.00
0753
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.02
0.00
0..00
"OT
2
0.04
0.10
0.03
' 0.01
0.00
0.01
0.00
0.18
1.56
0.14
,0.00
"2J07
0.00
0.01
0.00
0.00
0.00
0.00
0.00
0.01
0.05
0.01
0.00
'OF
3
0.27
0.35
0.11
0.02
0.00
0.05
0.00
0.47
3.60
0.41
0.00
5.28
0.02
0.02
0.01
0.00
0.00
0.00
0.00
0.03
0.10
0.01
0.00
"07211
Receptor Regions
456
0.14 0.06 0.35
0.23
0.18
0.14
0.14
0.06
0.04
0.15
1.21
6.92
0.03
9.23
0.01
0.02
0.01
0.02
0.03
0.00
0.00
0.01
0.05
0.79
.0...01
0.10
0.09
0.16
0.34
0.02
0.04
0.05
0.27
0.45
6.09
7.68
0.00
0.01
0.01
0.02
0.07
0.00
0.00
0.00
0.01
0.04
1.10
1725"
0.46
0.56
0.69
5.31
0.13
0.20
0.24
1.92
2.37
.0..05
12.26
0.02
0.03
0.03
0.08
1.05
0.01
0.01
0.02
0.09
0.28
0.01
1757
7
0.99
1.09
1.73
3.26
4.30
0.29
0.31
0.52
3.01
0.19
0.01
15.69
0.06
0.07
0.10
0.36
0.82
0.01
0.02
0.03
0.18
0.02
,0.00
1755"
8
1.26
1.94
7.10
24.73
0.67
0.57
1.14
0.76
1.23
0.04
0.0,0
3^741
0.07
0.11
0.41
2.98
0.12
0.03
0.05
0.05
0.08
0.00
D.,00
'375TJ
9
0.10
1.93
0.30
0.01
0.00
16.56
0.02
5.42
0.04
0.00
0..00
2O9
0.01
0.11
0.02
0.00
0.00
0.79
0.00
0.35
0.00
0.00
0.00
rrz9~
-------
10-15
Table 10-2.(continued).
Secondary 804
Receptor Regions
Source. .Regions
1
2
3
4
5
6
7
8
9
10
11
Total
Note: Totals
.Source, .Regions
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
A.
0.
1
01
02
00
00
00
00
00
05
09
00
.0.0
16
may not
2
0.02
0.05
0.02
0.00
0.00
0.01
0.00
0.09
0.15
0.03
0..0.0
0.35
agree
3
0.06
0.13
0.04
0.01
0.00
0.02
0.00
0.18
0.37
0.06
0.00
0.87
due to
0.
0.
0.
0.
0.
0.
0.
0.
0.
'0.
0.
1.
4
05
10
07
05
03
03
02
08
23
49
.01 ,
16
5
0.03
0.05
0.04
0.07
0.08
0.01
0.02
0.03
0.08
0.11
0.36
0.88
6
0.10
0.19
0.19
0.19
0.39
0.07
0.08
0.11
0.32
0.22
.0.01
1.88
7
0.21
0.36
0.40
0.38
0.26
0.12
0.09
0.21
0.41
0.04
.0.00
2.48
8
0.24
0.56
0.94
1.46
0.05
0.20
0.17
0.29
0.22
0.01
.Q.J30
9
0.04
0.42
0.10
o.oo
0.00
1.10
0.01
0.86
0.02
0.00
.0..00
1755
rounding.
Receptor.
.Regions
1 - Michigan
2 - Illinois-Indiana
3 - Ohio
4 - Pennsylvania
5 - New York to Maine
6 - Kentucky - Tennessee
7 - West Virginia to North Carolina
8 - Rest of Eastern U.S. (Florida
to Missouri to Minnesota)
9 - Ontario
10 - Quebec
11 - Atlantic Provinces
1 - Boundary waters
2 - Algoma
3 - Muskoka
4 - Quebec
5 - Southern Nova Scotia
6 - Vermont-New Hampshire
7 - Adirondacks
8 - Pennsylvania
9 - Smokies
-------
10-16
Table 10-3. Transfer Matrix for July 1978 Sulfur Dioxide and Sulfate
(Primary and Secondary) Concentrations (ug/m3) from the
ASTRAP Model Using the Phase II State/Provinces SC>2 Qnission
Inventory and Primary Sulfate Emission Factors in Table 10-1.
S02
Source Regions
1
2
3
4
5
6
7
8
9
10
11
Total
Primary 504
Source Regions
1
2
3
4
5
6
7
8
9
10
11
Total
1
0.00
0.02
0.00
0.00
0.00
0.00
0.00
0.06
0.37
0.00
0.00
7756"
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.02
0.00
0.00
"oTDT
2
0.05
0.15
0.02
0.00
0.00
0.00
0.00
0.18
1.03
0.00
0.00
"Ot
0.00
0.01
0.00
0.00
0.00
0.00
0.00
0.01
0.04
0.00
0.00
77DT
3
0.63
0.54
0.10
0.01
0.00
0.02
0.01
0.47
1.10
0.00
0.00
"278T
0.04
0.03
0.01
0.00
0.00
0.00
0.00
0.03
0.04
0.00
0.00
"oTTs
Receptor Regions
4-5 6
0.24 0.01 0.12
0.17
0.19
0.33
0.23
0.02
0.04
0.09
1.58
4.63
0.00
7738"
0.01
0.01
0.02
0.05
0.05
0.00
0.00
0.01
0.07
0.49
0.00
"Oo
0.02
0.06
0.21
0.39
0.01
0.05
0.01
0.16
0.33
3.40
1765
0.00
0.00
0.01
0.03
0.08
0.00
0.01
0.00
0.01
0.04
0.57
"0775
0.19
0.44
1.23
4.92
0.08
0.20
0.08
1.09
1.10
0.00
?75I
0.01
0.02
0.03
0.17
0.91
0.01
0.02
0.01
0.05
0.12
0.00
"OI
7
0.32
0.45
1.47
4.22
2.24
0.22
0.49
0.17
1.01
0.02
0.0,0
17760
0.02
0.03
0.09
0.42
0.41
0.01
0.03
0.12
0.06
0.00
0.00
T70S"
8
0.30
0.55
3.62
19.49
0.14
0.45
1.67
0.23
0.31
"o.oo
0.00
25776"
0.02
0.03
0.20
2.06
0.03
0.03
0.08
0.02
0.02
0.00
0..00
"OS"
9
0.00
0.07
0.03
0.00
0.00
12.52
0.01
3.39
0.00
0.00
0.0.0
lOT
0.00
0.00
0.00
0.00
0.00
0.54
0.00
0.22
0.00
0.00
0..00
7777
-------
Table 10-3.(continued).
10-17
Secondary 804
1
2
3
4
5
6
7
8
9
'10
11
Total
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
ft
00
02
00
00
00
00
00
06
18
00
00
27
' 0
0
0
0
0
0
0
0
0
0
0
ft
.04
.12
.02
.01
.00
.00
.01
.16
.19
.00
.00
75T
0.19
0.35
0.09
0.01
0.00
0.03
0.01
0.33
0.18
0.00
0..00
tn$
*. '
0.12
0.21
0.28
0.29
0.12
0.03
0.07
0.12
0.76
0.69
0.00
"2768"
0.02
0.03
0.11
0.24
0.24
0.01
0.11
0.02
0.18
0.22
0.39
trsfr
0.11
0.23
0.47
0.73
0.85
0.13
0.25
0.11'
0.57
0.21
0.00
TT£6
0.19
0.38
0.95
1.18
0.25
0.28
0.40
0.21
0.36
0.01
0.00
~%rn
0.16
0.40
1.45
2.34
0.03
0.45
0.79
0.27
0.11
0.00
0.00
"Oo"
0.00
0.04
0.03
0.00
0.00
1.72
0.01
0.94
0.00
0.00
,0.00
2T7T
Note: Totals may not agree due to rounding.
Source
1 - Michigan
2 - Illinois-Indiana
3 - Ohio
4 - Pennsylvania
5 - New York to Maine
6 - Kentucky - Tennessee
7 - West Virginia to North Carolina
8 - Rest of Eastern U.S. (Florida
to Missouri to Minnesota)
9 - Ontario
10 - Quebec
11 - Atlantic Provinces
r. -Regions
1 - Boundary waters
2 - Algoma
3 - Muskoka
4 - Quebec
5 - Southern Nova Scotia
6 - Vernont-New Hampshire
7 - Adirondacks
8 - Pennsylvania
9 - Smokies
-------
CHAPTER 11
CONCLUSIONS, RECOMMENDATIONS, AND PHASE III WORK PLANS
11.1 CONCLUSIONS AND RECOMMENDATIONS
The review of sulfur chemistry concluded that the rate of
homogeneous gas phase conversion of S02 to 304 is dominated by
free radical reaction processes and the concentration of the
important free radicals are dependent on many factors, espec-
ially the concentration of volatile organic compounds and
nitrogen oxides, the temperature and the solar radiation in-
tensity. Our knowledge of heterogeneous oxidation of S02 is
less complete, but indicates that liquid phase catalyzed oxida-
tion by manganese ion, carbon, and hydrogen peroxide all could
be potentially important. However, there is uncertainty about
the actual availability of these catalyzing substances in ambient
fine particulate matter. The review of nitrogen chemistry con-
cluded that the fate of nitric acid in the atmosphere is not well
understood, but it would still be useful to apply results of our
limited understanding of nitrogen chemistry to exploratory model-
ing exercises. It was concluded from these modeling exercises
that greater temporal and spatial resolutions for nitrogen model-
ing is necessary and more extensive and reliable monitoring in-
formation is required for validation purposes.
The review of the evidence for trends in precipitation com-
position and deposition concludes that, in spite of the difficul-
-------
11-2
ties with the data base and the controversy on the subject, the
the data do suggest expansion of the region covered by acidic
rainfall/ especially into the southeastern and mid-western
portions of the U.S. and the southeastern portions of Canada.
In this regard/ the Modeling Subgroup decided to focus its Phase
II efforts on the present situation and not to apply the models
to past data because of the uncertainties in past measurements
and other necessary input data, particularly the historic emiss-
ions fields.
The review of the seasonal variation of deposition and
chemical transformation rates concluded that many of the para-
meters in regional models could be strongly dependent upon
latitude during the winter months and recommended that not only'
seasonal variability, but also the spatial variability be taken
into account. This review provided some specific suggestions
for the participating models for Phase III.
The review of the global distribution of pH and its impli-
cations concluded that all precipitation appears to contain both
acidic and basic material in small quantities. There is evidence
that the "hemispheric background" value of precipitation pH (i.e.,
the average annual value at remote sites farthest removed from
the three major anthropogenic source regions in the northern
hemisphere) is significantly lower than the idealized "clean
atmosphere" value of 5.6. This average hemispheric precipitation
pH value may now be closer to 5.0 than to 5.6. More analysis
of data are required to determine what proportion of observed
-------
11-3
"background pH levels" are due to natural sources or to the
residual effect of man-made sources far upwind.
The Phase II Data Analysis review found the highest preci-
pitation acidity on an annual basis in the northern hemisphere
over (1) eastern North America, (2) western Europe, and (3)
Japan. Near neutral precipitation, frequently in excess of
pH 6, is found over the large continental areas of western
North America and Asia. The cause of slightly acidic precip-
itation along the west coast of North America has not yet been
completely explained, but may be due to either anthropogenic
sources or the release of biologically-produced organic sulfur
compounds from the Pacific Ocean surface, or both.
The zone of maximum acidity in eastern North America stretches
in a corridor through Ohio and Pennsylvania into Southern Ontario.
Available concentration data at remote locations in eastern North
America well removed from major source regions generally indicate
a summer sulfate maximum and a winter S02 maximum with highly
episodic behavior of both on a daily basis. In addition, cal-
culated dry depositions of sulfur are found to be of comparable
magnitude to wet sulfur depositions especially close to source
regions and in the winter season.
Recent interpretations of both concentration and deposition
monitoring data using trajectory calculations indicate that
maritime tropical air masses from the U.S. are the principal
conveyors of elevated concentrations and depositions to the
-------
11-4 '
extreme northeastern U.S. and southeastern Canada, as opposed to
continental polar air masses from Canada. It was recoginized in
making these interpretations that source-receptor relationships,
based upon calculations of transport and chemical transformations
between probable sources and the receptor of interest and upon
event data at single monitoring stations, are not always straight-
forward and are subject to uncertainty.
Finally, the data analysis review concluded that within east-
ern North America, natural sources of sulfur within the region
are unimportant compared to anthropogenic sources. Somewhat more
significant are the background sulfur concentrations that are trans-
ported to eastern North America from the Pacific and Carribean
Oceans, the Atlantic Ocean south of 30°N, and the arctic region.
These manifestations of the hemispheric background contribution
to acid deposition are still considered to be small in comparison
with the local and long-range transport impacts in eastern North
America. The predominant sources of elevated sulfur concentrations
in arctic air masses in the winter are thought to be, in order of
decreasing importance, Siberia, Europe, and Eastern North America.
The Phase II analysis of the role of modeling indicated that
model predictions are expected to deviate to some degree from
actual monitoring measurements. For practical reasons, models
do not incorporate all of our understanding of the relevant
physical processes, which itself is incomplete. Furthermore,
our available monitoring data bases are insufficient to compute
the ensemble average which the model is designed to predict. The
-------
11-5
uncertainties in model predictions may be quantified from the
differences between model predictions and observations.
Although the application of regional models is constrained
by these uncertainties inherent in their calculations, such
constraints can be alleviated to a significant degree by requir-
ing the modelers to quantify the relevant uncertainties, and by
taking these into account in any application. Some of the un-
certainties in the transfer matrix elements can be assessed by
analyzing the transfer matricies from more than one model and by
using probablistic techniques of analysis which will be developed
during Phase III. Other uncertainties may be quantified after
further model evaluation efforts are completed.
While there is still no general agreement in the modeling
community as to (1) the proper method and (2) the statistics for
intercomparison and evaluation of models, the Modeling Subgroup
made significant strides in selecting a common basis for perform-
ing these tasks for the eight participating models. In Phase II,
a complete set of evaluation statistics were computed by only
one modeler, while the monthly and annual residuals were computed
at 9 to 20 sites by three of the eight modelers. The other four
modelers are expected to complete this minimum evaluation by
September 1981. No single model has emerged as cleary superior
or inferior to the others from this first of three rounds of
model evaluation. The evaluation has primarily served to reveal
(1) the deficiences in the monitoring data bases, (2) the need
-------
11-6 '
for some changes in input parameters for some of the models,
and (3) the need to use at least one more year of independent
data for model evaluation.
The seasonal transfer matrices computed with several of
the models were too preliminary to draw any general conclusions/
but indicate that seasonal variability can be significant and
should be investigated further during Phase III. In addition,
most of the models completed detailed sensitivity analyses by
varying each input parameter separately within the range normally
used for long-range transport modeling with either (1) actual
meteorology and emissions or (2) a hypothetical source-receptor
situation and simulated meteorology. These sensitivity analyses,
which are documented in the individual Model Profiles, provide a
more complete understanding of the workings of'each model and
are useful for incorporating uncertainty in the analysis process.
The annual transfer matrices for Phases I and II were inter-
compared and, interestingly, the use of standardized inputs did
not reduce the range of variation among models in some of the
transfer matrix elements expressed in two of the three standard
forms, namely, absolute values or normalized by unit emissions
of sulfur. However, the transfer matrix elements expressed as
percentage contribution from a source region to a receptor area
were generally in much better agreement among the models. Since
additional refinements will be made to most of the models and
a new set of source-receptor regions will be used during Phase
III, it is premature to draw any general conclusions at this
-------
11-7 .
time. These variations in transfer matrix coefficients reflect
the current uncertainties in how best to parameterize all the
physical processes and are the result of different approaches by
independent modelers. Although the desirability of a single/
unified transfer matrix was recognized, the Modeling Subgroup has
reservations about the generation and application of a unified
transfer matrix at this time because no matter how it is generat-
ed its interpretation is subject to some question.
Since the modeling of nitrogen oxides chemistry is in its
infancy and because there is so little data available from which
to select the model parameterizations, any modeling at this stage
cannot yield more than some educated "first estimates" of the
long-range transport of nitrogen oxides. The preliminary results
from three models, in terms of transfer matrices and comparisons
to monitoring data, indicate the primitive nature of current
efforts. Therefore, at this stage, these should not be used
for analysis efforts.
A preliminary transfer matrix for primary sulfate emissions
was generated during Phase II to assess the relative contribu-
tion to acid deposition of primary sulfate emissions from oil-
fired and coal-fired combustion sources compared to that from
secondarily formed sulfates. Newly published primary sulfate
emission factors from large package boilers were utilized by
one regional transport modeler to develop this comparative
analysis, the analysis indicated that primary sulfate concen-
trations exceeded secondary sulfate concentrations in the winter
-------
11-8
season, while secondary sulfate concentrations exceeded primary
sulfate concentrations in all other areas in the winter and in
all areas during the summer. Additional modeling of the contribut-
ion of primary sulfate emissions will be performed during Phase III
using an improved emissions inventory and an attempt will be
made to evaluate the results against monitoring data.
-------
.11-9
11.2 PHASE III WORK PLAN
The principal objective for Phase III is to refine and
consolidate the information provided to the Coordinating
Committee at the end of Phase II. Some attention will be
directed to additonal transboundary air pollution issues that
are likely to be considered in the Air Quality Agreeemnt
negotiations.
The integrated Phase III report should then provide the
Coordinating Committee with substantially all the available
technical information and analysis relevant to closing nego-
tiations on a bilateral, transboundary air pollution agreement.
Before proceeding to a consideration of specific tasks
for Phase III, it is useful to review briefly the direction
provided by the Coordinating Committee and the Work Group 2
progress to the end of Phase II. The Coordinating Committee
recommended that Work Group 2 consider undertaking the following
tasks as early as possible in Phase II:
o Provide a means to estimate short-range and mesoscale
transport for sulfur compounds relative to long-
range transport for identified sensitive areas.
Provide a means for evaluating such transport, if
significant; and
o Assess the relative contribution to acid deposition
on identified sensitive areas of primary sulfate
emissions from oil-fired and coal-fired combustion
sources in comparison with secondary formed sulfate
from these sources. Compare the primary sulfate
deposition in identified sensitive areas from oil-
fired sources with the total sulfur depostion from
all other sources.
-------
11-10
The Coordinating Committee also requested that in addition
to completing the acid deposition analyses initiated in Phase
I, Phase III work programs analyze other important trans-
boundary air pollution issues. Among those issues of a
regional nature which are recognized to have an important
transboundary component are:
o Regional scale formation and transport of photochemical
oxidants
o Deposition and concentration of heavy metals and organics
In Phase II, Work Group 2 completed the following in
response to the above requests:
1. provided a list of mesoscale models;
2. estimated what proportion of the problem was due to
the short, the medium and the long-range transport
scales;
3. provided an initial evaluation of primary sulfate
emissions and their impact; and
4. provided some insight into the assessment iteration*
process
In Phase III (June 1981-January 1982), Work Group 2 will
do the following:
1. continue its evaluation and application of long-term
regional models and consolidate the results;
2. enlarge the modeling domain to include the entire
continental U.S. and western Canadian provinces;
* The Assessment iteration process involves repeatedly analyzing
transfer matrices, target depositions, and emissions and cost
vectors as they relate to possible control strategies.
-------
11-11
3. give additional emphasis to the atmospheric science
review and monitoring data analysis efforts;
4. run the selected (and other if appropriate) models
using the unified S02 and NOX emission inventories
on a state (multi-state)/province (sub-province)
basis and unify, if possible, the transfer matrices;
5. further quantify the "background S02 and NOX contribu-
tions" to the observed concentrations and depositions;
6. run the selected models on as many additional periods
of meteorological data as possible to assess the vari-
ability of the transfer coefficients (on a seasonal
and annual basis)insofar as possible;
7. propose a detailed work plan for the period beyond
Phase 3 on additional transboundary air pollution
issues as required by the Coordinating Committee;
8. address the issues raised by the peer reviews and
the assessment iteration process; and
To carry out its tasks in Phase III, the Working Group
has established the sub-group and activity coordination scheme
illustrated in Figure 11-1. Each sub-group will hold workshops
during Phase III.
An updated work schedule for Phase III is presented in
Figure 11-2.
-------
11-12
Figure 11-1. Phase III Organization for Work Group 2,
Co-Chairs
H. Ferguson
L. Machta
Vice Co-Chairs
G. Van Volkenburg
L. Smith
Co-Technical
Coordinators
J. Young
B. Niemann
1
Regional
Modeling
Sub-Group
Co-Coord inators
K. Demerjian
P. K. Misra
Monitoring and
Interpretation
Sub-Group
Co-Coord i nators
J. Miller
L. Barrie
Atmospheric
Science
Review
Sub-Group
Co-Coordinators
P. Altshuller
P. Sunimsrs
Local Source
Analysis
Co-Coord inators
To be selected
R. Shaw
-------
Figure 11-2. Proposed Work Group 2 Phase III Activity Schedule (Revised May 15, 1981).
ACTIVITY
1. REGIONAL MODELING SUB-GROUP
Id. Ifeciinlcdl pctii. tjL"Gup f&vitiW
JLU. MJilSO-LlQclCG tJ:anSJ.ejr JUauiXX WOiTK ^annUaJ./ SeaSOnaJ. ) »
la OtnGir pollutant nioctelincj 33 nGCGSSQiry cind f GcisiblG ~~
l Rsspona to PGGIT JTGVIGWS ~-~~ ~* _ . .
2. ATMOSPHERIC SCIENCE REVIEW SUB-GROUP
metals, etc.
2d. Review any new information on issues connected with Sulfur
and Nitrogen
2f. Analysis of initial comments on modeling, trends etc
9n Wfii-p enhni'Vrtin V£*rv»r-t- __J-_ _ __
July 81
x
"~~ X
X
X
A
Aug. 81
x
X
Sept 81
x
X
x
X
X
X
Oct. 81
X
x
Nov 81
Dec 81
x
X
X
X
Jan82
x
..
X
X
I
U)
-------
Figure 11-^^ontinued). Proposed Work Group 2 Phase III
Schedule (Revised May 15, 1981).
ACTIVITY
3. MONITORING INTERPRETATION SUBGROUP
3a. Continental background __ _
Jc. kJGdsonaJL variations - .... _«._.
JQ. L.iassi£ ication oy air mass origin etna episode cinuiyscis
3e. pH maps analysis ~ ~
3g. Analyze repot ts of anamolous or confounding roonitoring data ~
4. IJOCAL SOURCE ANALYSIS SUB-GROUP
deposition
to transboundary effects
analysis of data
*ie« wtiuc oui,>*yjroup j-trpoi. T. ~ - - . _«._«. ._- M _ «._._
5. W3RKING GROUP
Dti» Lxrcrput uuion or J.nucg.LUutX4 L ±nuj. i(jpoi.c -...,-.. . .- ....-...._-_ _ ...»,.«
July 81
X
X
Aug. 81
x
SeptS 1
X
X
X
X
X
X
X
X
X
Oct. 81
*
Nov. 81
X
Dec. 81
X
X
X
Jan82
»-
H-
H
.£
X
X
X
-------
Chapter 12
PRELiM.i.NAJR.Y,' _PfiQ.P.6.SA.LS '.FQR' RESEARCH., ' MQD.E.L.I.N.G., .AND.'
M.Q.NJ. TORI ' NG. ELEMENT. .OF. Tfl.E.
12.1 ,1 At ic od .u.c.t ,ip n
A transboundary air pollution agreement between the
United States and Canada that reflects our scientific under-
standing in the early 1980s of this problem may not be sufficient
to resolve completely all aspects of the problem. Therefore,
it is important to provide a mechanism whereby new knowledge
can be acquired and assimilated by those responsible for
managing the environmental protection programs of the two
countries. Additionally, recognizing our limitations to
predict with certainty the full consequences of controlling
emissions, it is important to determine by observations of
concentrations and deposition rates the effectiveness of the
initial agreement in resolving transboundary impact issues.
The following four subsections suggest areas where uni-
lateral and cooperative programs aimed at providing increased
understanding could further this objective. They have been
categorized under two headings. First, E,s,s,en.t,ial items are
those that are considered to be necessary for improved
understanding in the near term (next .few years). It is
recommended that the programs of both governments include
sufficient resources to adequately fund these activities.
Second, .Desirable items are those that are considered useful
-------
12-2
for improved understanding in the short-term/ or may be
essential to the long-term resolution of some transboundary
issues. In making the following recommendations the Work
Group has taken into consideration current U.S. and Canadian
planning information as embodied in the U.S. National Acid
Precipitation Assessment Plan and the Canadian LRTAP plan.
The Work Group recommends that a bilateral air quality res-
earch and monitoring committee be established to coordinate
these activities beyond January, 1982.
12.2 Atmospheric Processes Research
Essential
0 scavenging mechanisms for wet and dry removal of
transported pollutants.
0 background concentrations and ventilation losses
at ground level and aloft.
0 atmospheric chemistry for acid formation and
neutralization processes, especially those for
nitrogen chemistry
0 improved understanding of long-distance meteorological
transport.
0 prediction of meteorological conditions associated
with transboundary air pollution.
Desirable
0 reconstruction of transport and dispersion during
episodes
-------
12-3
0 global background of acids and precursors concentra-
tion/ extent, frequency and their explanation.
0 cooperative tracer studies.
0 atmospheric chemistry of oxidant precursors and their
aged reaction products.
0 vertical sulfur concentrations and regional horizontal
flux studies.
12.3 Monitoring
Essential
0 development of improved acid deposition monitoring
devices, particularly for dry deposition.
0 development of uniform protocol for acid deposition
monitoring procedures, sample handling and analysis,
data archiving, quality assurance, and exchange of
data sets between the US and Canada.
0 coordination of gaseous and particulate pollutant
monitoring for establishing air quality trends.
0 sustained monitoring program for acid deposition and
air pollutant concentrations in North America for:
remote sites potentially affected and sensitive
areas currently impacted.
along the Canada/U.S. border.
0 monitoring fluxes crossing the periphery of the
modeling region.
-------
12-4
Desirable
0 . feasibility study on the establishment of cooperative
episode warning network for harmful pollutants.
0 assessment of needs for, and as appropriate/ imple-
mentation of,
- a heavy metals monitoring network.
- an oxidants monitoring network.
a toxic organics monitoring network.
a supplementary visibility monitoring network.
12.4 Modeling
Essential
0 study approaches for providing improved quantification
of the uncertainties associated with modeling source-
receptor relationships.
0 continue validation and application of available
models to the North American situation so as to
provide increased understanding of source-receptor
relationships.
0 development of a comprehensive North American acid
deposition model.
Desirable
0 development of a regional oxidant model for application
in eastern North America.
0 development of regional haze model for application
in eastern North America.
-------
12-5
0 development of a heavy metals atmospheric transport
model.
0 development of a toxic organics atmospheric transport
model.
0 development of techniques for predicting the occurrence
in real time of transboundary air pollution episodes.
12.5 Atmospheric Science Assessment
Essential
0 produce a regular(annual) bilateral report which updates
our current understanding of transboundary air pollution
phenomena, including our ability to monitor it and model
its source-receptor relationships. (A continuing bi-
lateral committee could be charged with this task.)
0 produce critical reviews/ as warranted, of major
developments affecting our understanding of trans-
boundary air pollution.
-------
R-l
REFERENCES
Barrie, L. A., et al., 1980: The Canadian Air and Precipitation
Monitoring Network APN, in Atmospheric Pollution 1980, Proceedings
of the 14th International Colloquium, Paris, France, May 5-8,
Studies in Environmental Science, Volume 8, Elsevier, Amsterdam.
Barrie, L. A., 1981: Environment Canada's Long Range Transport of
Atmospheric Pollutants Program: Atmospheric Studies, Paper
Presented at the Conference on the Effects of Acid Precipitation
On Ecological Systems in the Region of the Great Lakes, March 31-
April 3, Michigan State University.
Barrie, L. A., 1980: The Fate of Particulate Emissions from An
Isolated Power Plant in the Oil Sands Area of Western Canada,
Annals. New York Acad. Sci., 338, 434-452.
Barrie, L. A., et al., 1981: The Influence of Mid-Latitudinal
Pollution Sources on Haze in the Canadian Arctic, Atmos.
Environ., 15, in press.
Barrie, L. A., et al., 1981: The Temporal and Geographical
Distribution of Acidic Pollutants in the Atmosphere of Eastern
Canada, Proc. Conf. on Long-Range Transport of Airborne Pollutants,
April 29-May 1, Albany, New York.
Barrie, L. A., 1981: The Prediction of Rain Acidity and S02
Scavenging in Eastern North America, Atmos. Environ, 15, 31.
Beilke, S., 1970: Laboratory Investigations of Washout of
Trace Gases, AEC Symp. Ser. Proc. Symp. on Precip.
Scavenging.
Benkeley, C. W. and Mills, M. T. 1980: Development and Validation
of a Simple Regional Climatological Dispersion Model for Nitrogen
Oxides, Report Prepared for Argonne National Laboratories by
Teknekron Research, Inc., November.
Benkowitz, C., 1979: Compiling a Multistate Emissions Inventory,
Brookhaven National Laboratory/MAP3S Report, BNL-26843,
Brookhaven National Laboratory, Upton, N.Y.
-------
R-2
Bolin, B. and Persson, C., 1975: Regional Dispersion and
Deposition of Atmospheric Pollutants with Particular Applica-
tion to Sulfur Pollution over Western Europe, Tellus, 27,
281-310.
Bottger, et al., 1978. Atmospharische Kreislaufe van
Strichoxiden and Ammoniak, Kernforschungsaulage Julich GmbH,
Inst. Chemie: Jul-1558.
Botthheim, J. W. , 1981: Discussion on the Incorporation of
Oxides of Nitrogen Chemistry in LTRAP Models, Atmoshperic
Environment Service Internal Report.
Chung, Y. S., 1977: Sources and Sinks of Photochemical Ozone
(03) in the Boundary Layer, Proceedings of the Fourth
International Clean Air Congress, Tokoyo, 138.
Clark, T. L., 1980: Annual Anthropogenic Pollutant Emissions in
the United States and Southern Canada East of the Rocky
Mountains, Atmos. Environ., 14, 961-970.
Reisinger, L.M. and Crawford, T. L., 1981: Interregional
Transport: Case Studies of Measurements Versus Model Predictions,
Paper 81-13.4 presented at the 74th Annual Meeting of the APCA,
June 21-26, Philadelphia, PA.
Fay, J. A. and Rosenzweig, J. J. 1980: An Analytical Diffusion
Model for Long Distance Transport of Air Pollutants, Atmos.
Environ., 14, 355-365.
Galloway, J. N., and Whelpdale, D. M, 1980: An Atmospheric
Sulfur Budget for Eastern North American, Atmos. Environ., 14,
409-417.
Gravenhorst, G., et al., 1980: Sulfur Dioxide Absorbed in Rain
Water, in Proceedings of the Effects of Acid Precipitation on
Terrestrial Ecosystms, Plenum Press, New York.
Hales, J. M. 1981: OSCAR and Miscellany, Directions Memo, MAP3S/
RAINE Program, June 1, 6 pp.
Henry, R.C. and Hidy, G. M. , 1980: Potential for Atmospheric
Sulfur from Microbiological Sulfate Reduction, Atmos Environ.,
14, 1095-1104.
-------
R-3
Hicks, B.B. and Shannon, J. D. , 1979: A Method for Modeling the
Deposition of Sulfur by Precipitation Over Regional Scales,
.J,./ Appl,.; M.e.teror..' , X8, 1415-1420.
Homolya, J. B. and Lambert, S. 1981: A Charactrization of
Sulfate Emissions from Non-Utility Boilers in New York City Firing
Low-Sulfur Residual Oils, .Ji_\Air."_jPQ,ll.|,"Qqp,t,j_.'_Aggp.c_ i 31,
139-143.
Huschke, R. E., 1969: Arctic Cloud Statistics from 'Air
Calibrated1 Surface Weather Observations, U.S. Air Force
Project, Rand Contract No. F 44620-67-C-0015.
Jeffries, H. E., 1979: Use of Laboratory Data to Describe
the Fate of NOX in the Atmosphere, Presented at the Technical
Symposium on the Implications of a Low NOX Vehicle Emission
Standard, Reston, Virginia, May 2-4, 1979.
Johnson, W. B., et al., 1978. Long-Term Regional Patterns and
Transfrentier Exchanges of Airborne Sulfur Pollution in Europe,
.A.tmpS,..'.E.nyi.rqn.yf .1.2, 511-527.
Koerner, R, M. and Fishe,D. 1981: Acid Snow in the Canadian
High Arctic, submitted to Nature,
Lamb, R. G., 1980: Mathemetical Principles of Turbulent Diffusion
Modeling, Article in Atmospheric Planetary Boundary Layer
Physics, Elsevier Publishing, Amsterdam.
Levine, S. Z. and Schwartz, S. E, 1981. In-cloud and Below-
Cloud Scavenging of Nitric Acid Vapor, A.tmp,s.., Eny.i.rpn. ,
15, in press.
Lyons, W. A., 1980: Evidence of Transport of Hazy Air Masses
from Satellite Imagery,' Annal,., N..Y... A.c.ad... S.cj... , .33,8, 418-433.
MAP3S/RAINE, 1981: The MAP3S/RAINE Precipitation Chemistry
Network: Statistical Overview for the Period 1976-1980,
submitted to Atmp.s.y' .Eny.iy.on
Merritt, W. F, 1976: Trace Element Content of Precipitation
in a Remote Area, in Measurement.,, Pe.te.cti.oa^ a^nd, .Cpntrp.!
p_f._,E.ny.i.r.o.nm.e.n.t.a.l,'_P.Q 11 At.an.ts, Internationa 1 AtomTc Energy
Agency 7 Vienna7 7 5-87.
-------
R-4
Mueller, P. K., et al., 1979: Some Early Results from the
Sulfate Regional Experiment (SURE), in Proc. of the 4th
Symposium on Turbulence, Diffusion and Air Pollution, January
15-18, Reno, Nevada, American Meterological Society, Boston, MA.
Niemann, B. L., et al., 1980: Initial Evaluation of Regional
Transport and Subregional Dispersion Models for Sulfur
Dioxide and Fine Particulates, Proceedings of the Second
Joint AMS/APCA Conference on Applications of Air Pollution
Meteorology, March 24-27, New Orleans, LA.
Niemann, B. L., et al., 1979: Application of a'Regional Trans-
port Model to the Simulation of Multi-Scale Sulfate Episodes
Over the Eastern United States and Canada, in Proceedings of
the WMO Symposium on the Long-Range Transport of Pollutants
and Its Relation to General Circulation Including Stratospheric/
Troposheric Exchange Processes, October 1-5, Sofia, Bulgaria.
Olson, M. P., et al. 1979: A Concentation/Deposition Model
Applied to the Canadian Long-Range Transport of Air Pollutants
Project, LRTAP 79-5, Atmospheric Environment Service, Downsview,
Ontario.
Paresh, P. P. and Husain L., 1981: Windflow Patterns and
Particulate Sulfate Concentration in Ambient Air at Whiteface
Mountain, New York: A Continuous 18 Month Inventory, submitted
to Science.
Patterson, D. E., et al. 1981: Monte Carlo Simulation of Daily
Regional Sulfur Distribution: Comparison with SURE Sulfate
Data and Visual Range Observations During August 1977, j,..'
APPl... M.e.t.epf,* , ,20, 70-86
Peterssen, S. 1956: .Weather, .Analysis, .ap.d. fpre.c.as.tipg, McGraw-
Hill, New York.
Portelli, R. V. , 1977: Mixing Heights, Wind Speeds and Ventilation
Coefficients for Canada, Climatological Studies No. 31,
Atmospheric Environment, Downsview, Canada.
Rutherford, I. D., 1977: An Operational Three Dimensional
Multivariate Statistical Objective Analysis Scheme, Issue #1,
Notes Scientifiques at Techniques, RPN, Atmospheric Environment
Service, Dorval, PQ.
Samson, P. J., 1980: Trajectory Analysis of Summertime Sulfate
Concentrations in the Northeastern United States. ,»J..__.'_Appl..
Meteor-, 1.9, 1382-1394.
-------
R-5
Shannon, J. D., 1979: The Advanced Statistical Trajectory
Regional Air Pollution Model, in Proceedings of the Fourth
Symposium on Turbulence, Diffusion, and Air Pollution, January
15-18, Reno, Nevada, American Meteorological Society,
Boston, MA.
Shannon, J.D., 1981: A Model of Regional Long-Term Average
Sulfur Atmospheric Pollution, Surface Removal, and Net
Horizontal Flux, Atmos. Environ, 15, 689-701.
Shieh, C. M., et al., 1979: Estimated Dry Deposition Velocities
of Sulfur Over the Eastern United States and Surrounding Regions,
Atmos. Environ., 13, 1361-1368.
Shenfeld, L. , et al., 1980: An Air Pollution Incident in Southern
Ontario, Canada with International Implications, Ontario
Ministry of the Environment, Toronto, Ontario.
Slinn, W. G. N., et al., 1979: Wet and Dry and Resuspension of
AFCT/ TFCT Fuel Processing Radionuclides, U.S. Department of
Energy Final Report SR-0980-10, Air Resources Center, Oregon
State University, Corvallis, Oregon.
Sykes, R. I. and Hatton, L. , 1976: Computation of Horizontal
' Trajectories Based on the Surface Geotrpphic Wind, Atmos.
Environ., 10, 925.
Szabo, M. F., et al., 1981: Perspectives on the Issue of Acid
Rain, Final Draft Report to the U.S. Department of Energy,
Morgantown Energy Technology Center, Contract No. DE-AC21-
81MC16361, June.
Tennekes, H., 1977: The General Circulation of Two-Dimensional
Turbulent Flow on a Beta plane, J. Atmos. Sci., 34, 702-712.
Voldner, E. C., et al., 1980: A Preliminary Canadian Emissions
Inventory for Sulfur and Nitrogen Oxides, Atmos. Environ., 14,
419-428.
Voldner, E. C., et al., 1980: Comparison Between Measured and
Computed Concentrations of Sulfur Compounds in Eastern North
America. AQRB-80-0003-T (LRTAP-02), Atmospheric Environment
Service, Downsview, Ontario (To be published in Journal of
Geophysical Research, 1981).
-------
R-6
Walmsley, J. L., et al., 1981: Sensitivity Tests with a Trajectory
Model/ to be published in the Air Quality Research Branch
Report Series, Atmospheric Environment Service, Downsview.
Whelpdale, I. M. 1978: *jaiT9e~?9a-'-e Atmospheric Sulfur Studies
in Canada, Atpips...'Ep.v;..rpp.., .1,2, 661-670.
Wilson, J. et al., 1980: Wet Deposition in the Northeastern
United States, Atmospheric Sciences Research Center Publication
796, State University of New York at Albany, December.
Wilson, W. E/ and Gillani, N. V., 1979: Transformation During
Transport: A State-of-the-Art Survey of the Conversion of S02 to
Sulfate, WMO Symposium on the Long-Range Transport of Pollutants
and Its Relation to General Circulation Including Stratospheric/
Tropospheric Exchange Processes, 1-5 October, Sofia, Bulgaria.
-------
ADDENDUM A
Model Critique by L. Machta
U.S. Co-chairman of Work Group 2
9
The models in this report which provide the transfer
matrices represent the cutting edge of the state-of-the-art
in acid rain modelling. They generally incorporate in a
practical way the best available physics, chemistry and
meteorology. Their products have been tested/ insofar as
possible/ against many real atmosphere measurements. Thus,
the criticism below is not directed at deliberate misconceptions
or imperfections of acid rain models, but rather at the current
inadequate state-of-the-art.
It is this writer's judgment that the acid rain models
are not yet good enough to use their transfer matrices to
play a role in assessing emission controls to reduce acid
rain impacts. The reasons for this inadequacy are noted below:
1. Current acid rain models are currently-capable of
predicting only the sulfate and SC>2 deposition and air concen-
trations. They cannot, as yet, directly predict deposition
or concentration of hydrogen ions, NOX, NO-j", and other major
ions like ammonia, calcium or magnesium that bear on acid rain.
2. The models do not account for the low pH in isolated
regions of the globe where the precipitation often is almost
(or just) as acid as downwind of large anthropogenic sources.
-------
A-2
The presence of this global acid rain means either that there
are larger natural sources or longer distance transport from
man-made sources than implied by the present models.
3. The flux to the ground, vegetation, water, etc.
cannot now be adequately measured. This leaves the budgets
of acid material in doubt and precludes complete model valida-
tion. It also prevents a correlation between total acid
deposition and the observed impact in the field.
While the above represent the writer's main reservations,
some critics have noted other factors which also possess
merit such as: the sulfate chemistry in the models which is
simplified may not adequately represent real atmospheric
chemistry; transport and dispersion calculations over long
distances have not been validated; and precipitation scavenging
is also simplified based on very limited data.
To summarize, the criticism of models applies mainly to
the inadequate state of knowledge rather than the model's
incorporation of known information. This writer does not
believe that they should be used as yet to assess the benefits
of emission restrictions. The above, however, should not be
interpreted as contending that no benefit will accrue from
fewer emissions of man-made sulfur oxides, but rather that the
models are not yet ready to quantitatively predict the reduc-
tion in acid deposition.
-------
ADDENDUM B
Response to Dr. Machta's Addendum A
by H. L. Ferguson
Canadian Co-Chairman of Work Group 2 <
Dr. Machta's critique identifies some aspects of long
range transport modeling where virtually all modelers
recognize that our knowledge is incomplete. Much of the
current and planned activity of Work Group 2 is aimed at
reducing the uncertainties he describes.
The basic question of whether or not the transfer matrices
are good enough to "play a role in assessing emission controls
to reduce acid rain impacts" is a matter c?f "shades of grey"
judgment. At what_point does such a synthesis of scientific
model estimates (with their implied error bands) become "good
enough"? Canadian modelers generally hold the view that the
transfer matrices provide sufficiently indicative information
linking causes-and effects between larger areas (such as
groups of states) to be used for implementing initial control
actions. This view appears to be shared by many European and
U.S. modelers.
Some general philosophical comments on the present stage
of LRTAP models and their potential applications are presented
in Chapter 4 of this report.
With reference to the three specific points raised by
Dr. Machta, I would like to make the following observations:
-------
B-2
1. The capability of estimating hydrogen ion concentra-
ion is particulary important because of its direct
relationship with pH. Models can predict deposition
and concentrations of hydrogen ions indirectly using
various methods. The sulfate surrogate method/ for
example, reproduces very well the H+ concentration
in Eastern North America (Barrie, 1981).
Preliminary linear nitrogen and nitrate modeling
is presented in Chapter 9 of this report and is
discussed by several of the modelers.
Modeling of other major ions has not been attempted
to date because the required inventories do not
»
exist.
2. Models have not been designed for and have not been
applied to the prediction of pH in isolated regions
of the globe, so we cannot conclude anything about
our skill in predicting in these areas. We do know
that in such areas the data base is grossly inade-
quate for developing and testing LRT models. We
also know that the "low" average precipitation pH
values being reported at a few remote locations still
represent 5 to 10 times less hydrogen ion than average
values in the most affected area of eastern North
America.
Very long range transport does occur and has been
widely reported. A few examples are:
(1) Large uncontrolled forest fires in Western
-------
B-3
North America in the early 1950's caused a
blue haze over Europe for many weeks;
(2) Recent measurements coupled with back trajec-
tories in the Arctic have indicated sources in
i
the USSR, China and Eastern North America
(Barrie, 1981) ; and
(3) Preliminary analysis of the Bermuda wet
chemistry data indicate that low pH episodes
have their origin in Eastern North America
while higher pH episodes are associated with
long sea trajectories from the south and east
(personal communication from John Miller).
3.* All flux components (wet and dry) can and have been
measured although the error bands are large in some
cases. Dry flux is not, however, measured routinely
and is, therefore, not available for validation
purposes.
Intercountry sulphur budgets have been independently
determined by various authors using various techniques,
in both the USA and Canada, and are generally in good
agreement.
The Working Group 1, Phase II data show an excellent
correlation between measured excess sulfate in lake
and river systems and modeled wet sulfate deposition
from the models, indicating that only a small fraction
-------
B-4
of the dry deposited sulphur becomes free hydrogen
ion in an aquatic system. This allows modeled
results to be tied directly to effects in the
t aquatic ecosystem.
Plans call for a high priority to be given to more
integration of the Working Group 1 and 2 products
in Phase III.
We accept that there is uncertainty in all measured and
modeled parameters. I think we can better define the limits
of confidence during Phase III. There is no point, however,
in improving the chemistry, or any other parameterization, to
produce changes in output that are insignificant compared to
the errors in the measured data.
I do not personally believe that the models are misre-
presenting the major features of the acid deposition problem
in eastern North America.
A comparison of available hemispheric data on acid
depositions with major hemispheric emission regions of acid
precursors shows that sensitive pristine regions are being
adversely affected by source regions hundreds to thousands of
kilometers distant.
Current models and transfer matrices show the general
source-receptor relationships involved. They represent the
best method available for evaluating the potential ameliorative
effects of alternative control strategy scenarios.
-------
B-5
effects of alternative control .strategy scenarios. In my
view they provide sufficient information to make a start now
on a broad agreement for reversing the deterioration of the
environment due to acid deposition in Eastern North America.
As the models and transfer matrices continue to be improved
through substantial longer term research efforts in the
United States/ Canada and elsewhere, the initial broad agree-
ment on controls can and should be periodically reviewed and
refined.
-------
APPENDIX 1
Work Group 2
Terras of Reference
and Additional Guidance
-------
A.1-1
Terms of Reference from the MOI
The Group will provide information based on cooperative
atmospheric modeling activities leading to an understanding
of the transport of air pollutants between source regions and
sensitive areas, and prepare proposals for the "Research,
Modeling and Monitoring" element of an agreement. As a first
priority the Group will by October 1, 1980 provide initial
guidance on suitable atmospheric transport models to be used
in preliminary assessment activities.
In carrying out its work, the Group will:*
identify source regions and applicable emission
data bases;
evaluate and select atmospheric transport models
and data bases to be used;
relate emissions from the source regions to
loadings in each identified sensitive area;
- calculate emission reductions required from source
regions to achieve proposed reductions in air
pollutant concentration and deposition rates which
would be necessary in order to protect sensitive
areas;
* proposed additional term of reference:
11 - evaluate and employ available field measurements,
monitoring data and other information;"
-------
B-4
of the dry deposited sulphur becomes free hydrogen
ion in an aquatic system. This allows modeled
results to be tied directly to effects in the
+ aquatic ecosystem.
Plans call for a high priority to be given to more
integration of the Working Group 1 and 2 products
in Phase III.
We accept that there is uncertainty in all measured and
modeled parameters. I think we can better define the limits
of confidence during Phase III. There is no point, however,
in improving the chemistry, or any other parameterization, to
produce changes in output that are insignificant compared to
the errors in the measured data.
I do not personally believe that the models are misre-
presenting the major features of the acid deposition problem
in eastern North America.
A comparison of available hemispheric data on acid
depositions with major hemispheric emission regions of acid
precursors shows that sensitive pristine regions are being
adversely affected by source regions hundreds to thousands of
kilometers distant.
Current models and transfer matrices show the general
source-receptor relationships involved. They represent the
best method available for evaluating the potential ameliorative
effects of alternative control strategy scenarios.
-------
B-5
effects of alternative control strategy scenarios. In my
view they provide sufficient information to make a start now
on a broad agreement for reversing the deterioration of the
environment due to acid deposition in Eastern North America.
As the models and transfer matrices continue to be improved
through substantial longer term research efforts in the
United States, Canada and elsewhere, the initial broad agree-
ment on controls can and should be periodically reviewed and
refined.
-------
APPENDIX 1
Work Group 2
Terras of Reference
and Additional Guidance
-------
A.1-1
Terms of Reference from the MOI
The Group will provide information based on cooperative
atmospheric modeling activities leading to an understanding
of the transport of air pollutants between source regions and
sensitive areas, and prepare proposals for the "Research,
Modeling and Monitoring" element of an agreement. As a first
priority the Group will by October 1, 1980 provide initial
guidance on suitable atmospheric transport models to be used
in preliminary assessment activities.
In carrying out its work, the Group will:*
identify source regions and applicable emission
data bases;^
evaluate and select atmospheric transport models
and data bases to be used;
- relate emissions from the source regions to
loadings in each identified sensitive area;
calculate emission reductions required from source
regions to achieve proposed reductions in air
pollutant concentration and deposition rates which
would be necessary in order to protect sensitive
areas;
* proposed additional term of reference:
" - evaluate and employ available field measurements,
monitoring data and other information;"
-------
A.1-2
- assess historic trends of emissions, ambient
concentrations and atmospheric deposition to gain
further insights into source-receptor relationships
for air quality, including deposition; and
prepare proposals for the "Research, Modeling and
Monitoring"-element of an agreement.
Additional Guidance from the Chairman of WG 3A
Each Work Group will be responsible individually for the
following:
a. Develop data needs and analysis methods for their Work
Group; identify required inputs from other Work Groups;
(due to the size of the Work Groups, the Chairmen will
have to very carefully orchestrate the Group's activities
in order to accomplish their tasks).
b. The technical review (including peer review as necessary)
of their work products.
c. Maintaining agreed upon work schedules with prompt
notification to 3A Chairman in the event of any
significant deviation from Work Plan.
d. Responsible for coordination with their counterparts
from the other country in conducting full cooperative
analyses in order to fulfill the terms of reference.
e. Responsible for fulfilling requests for information
from other work groups in a timely fashion.
-------
A.1-3
f. Be prepared to draft language for portion of agreement
that pertains to their tasks as directed by Coordinating
Committee.
-------
APPENDIX 2
Membership of Work Group 2
-------
1. United States
A.2-1
Chairman:
Vice Chairman;
Lester Machta, Director
Air Resources Laboratory, (Room 613)
National Oceanic and Atmospheric
Administration
8060 13th Street
Silver Spring, MD 20910
(301) 427-7645
Lowell Smith, Director
Program Integration and
Policy Staff (RD-681)
Environmental Protection Agency
Washington, D. C. 20460
(202) 426-9434
Members:
Paul Altshuller
Environmental Sciences Research
Laboratory (MD-59)
Environmental Protection Agency
Research Triangle Park, NC 27711
(919) 629-2191
Franz Burmann
Environmental Monitoring Systems
Laboratory (MD-75)
Environmental Protection Agency
Research Triangle Park, NC 27711
(919) 629-2106
Richard Harrington *
Morgantown Energy Technology Center
Department of Energy
Morgantown, West Virginia 26505
(304) 599-7529
Roger Morris
Office of Policy and Evaluation (PE-83)
Department of Energy
1000 Independence Avenue, S.W.
Washington, D. C. 20585
(202) 252-6453
-------
A.2-2
Bernard Silverman
Water and Power Resources Services
E & R Center P. 0. Box 25007
Department of Interior
Bldg. 67 - Denver Federal Center
Denver, CO 80225
(303) 234-2576
Alternate for Silverman
Richard Ives
Department of Interior, Code 124
Washington, D.C. 20240
(202) 343-6703
John Ficke *
Council of Environmental Quality
722 Jackson Place, N. W.
Washington, D. C. 20006
(202) 395-5760
Richard Ball *
Regional Impacts Division
Department of Energy (EV-24)
Washington, D. C. 20545
(301) 353-5801
Ken Demerjian
Meteorology and Assessment Division (MD-80)
Environmental Protection Agency
Research Triange Park, NC 27711
(919) 629-3660
Nels Laulainen *
Office of Environmental Processes
and Effects Research (RD-682)
Environmental Protection Agency
Washington, DC 20460
(202) 426-0803
Brand Niemann
Program Integration and Policy Staff (RD-681)
Environmental Protection Agency
Washington, DC 20460
(202) 755-0324
Joe Tikvart
Office of Quality Planning and Standards (MD-14)
Environmental Protection Agency
Research Triangle Park, NC 27711
(919) 629-5261
-------
A.2-3
Terry Clark *
Meteorology and Assessment Division (MD-80)
Environmental Protection Agency
Research Triangle Park, N.C. 27711
(919) 629-4524
John Miller
Air Resources Laboratory
National Oceanic and Atmospheric
Administration
8060 13th Street
Silver Spring, MD. 20910
(301) 427-7645
Jack Blanchard
OES/ENH Room 7820
State Department
2101 C Street, N. W.
Washington, DC 20520
(202) 632-5748
Liaison: Robin Porter
Department of State
EUR/CAN Room 5227
2101 C Street, N.W.
Washington, DC 20520
(202) 632-3189
Dolores Gregory
Office of International Activities (A-106)
Environmental Protection Agency
Washington, DC 20460
(202) 755-0430
* Indicates new member of the Work Group
-------
2. Canada
A.2-4
Chairman:
Vice Chairman;
Howard Ferguson, Director
Air Quality and Inter-environmental
Research Branch
Atmospheric Environment Service
4905 Dufferin Street
Downsview, Ontario M3H5T4
(416) 667-4937
Greg Van Volkenburgh, Director
Air Resources Branch
Ontario Ministry of the Environment
880 Bay Street, 4th Floor
Toronto, Ontario, M5S1Z8
(416) 965-6343
Members:
Douglas M. Whelpdale
Air Quality and Inter-environmental
Research Branch
Atmospheric Environment Service
Environment Canada
4905 Dufferin Street
Downsview, Ontario, M3H5T4
(416) 667-4785
James W.S. Young
Air Quality and Inter-environmental
Research Branch
Atmospheric Environment Service
Environment Canada
4905 Dufferin Street
Downsview, Ontario, M3H5T4
(416) 667-4786
Marvin P. Olson
Air Quality and Inter-environmental
Research Branch
Atmospheric Environment Service
Environment Canada
4905 Dufferin Street
Downsview, Ontario M3H5T4
(416) 667-4903
Peter W. Summers
Air Quality and Inter-environmental
Research Branch
Atmospheric Environment Service
Environment Canada
4905 Dufferin Street
Downsview, Ontario, M3H5T4
(416) 667-4785
-------
A.2-5
Frank Vena *
Pollution Data Analysis Division
Environmental Protection Service
Environment Canada
Place Vincent Massey
Ottawa, Ontario, K1A1C8
(819) 997-3354
B . Power
Environmental Management
and Control Division
Newfoundland Department of Provincial
Affairs and Environment
Elizabeth Towers
St. John's, Newfoundland
G. Paulin
Directeur,
Director de la Recherche
Ministere de 1 'Environnement ,
2360 Chemin Ste-Foy
Quebec G1V 4H2
(418) 643-2073
P. K. Misra *
Air Quality and Meteorology Section
Air Resources Branch
Ontario Ministry of Environment
880 Bay Street, 4th Floor
Toronto, Ontario, M5S1Z8
(416) 965-5068
Liaison: R. Beaulieu
United States Transboundary
Relations Division
Department of External Affairs
125 Sussex Drive
Ottawa, Ontario, K1AOG2
(613) 996-6620
Hans Martin
LRTAP Liaison Office
Atmospheric Environment Service
4905 Dufferin Street
Downsview, Ontario, M3H5T4
(416) 667-4824
Indicates new member of the Work Group
-------
Appendix 3
Glossary of Terms
-------
A..3-1
Introductory Comments
During the preparation of this glossary, use has been
made of terminology and definitions found in, inter alia, the
first two annual reports of the United States-Canada Research
Consultation Group on the Long Range Transport of Air Pollutants,
and the draft Federal Acid Rain Assessment Plan. An obvious
need exists for uniformity in terminology amongst all Work
Groups and others involved in activities related to the
Memorandum of Intent and subsequent developments. It is
anticipated that this glossary will grow and be refined as
further contributions from specialists in various disciplines
are received.
-------
A.3-2
Acid Deposition; Collectively, the processes by which acidic
and acidifying materials are removed from the atmosphere and
deposited at the surface of the earth. Also, the amount of
material so deposited. (Units: ML~2T"~1.)
Acid Precipitation; A more precise term than acid rain, it
usually refers to all types of precipitation with pH less
than 5.6.
Acid Rain; A popular term used to describe precipitation that
is more acidic than "clean" rain (pH 5.6). It is also used
more generally to describe other atmospheric depositipn
phenomena involving acidity.
Analytical Model; A mathematical model in which the solution
to the system of governing equations is expressed in terms of
analytical functions. As such, these models are simplifications
of Lagrangian, Eulerian or statistical models.
Anthropogenic; Produced by man's activity.
Bulk Deposition; The term applied to atmospheric deposition
collected in a collector which is open at all times. Bulk
deposition consists of wet deposition, plus an unknown fraction
of the dry particulate deposition, plus an unknown and probably
very small fraction of the dry gaseous deposition.
Dry Deposition; Collectively, the processes, excluding preci-
pitation processes, by which materials are removed from the
atmosphere and deposited at the surface of the earth. Processes
include sedimentation of large particles, the turbulent transfer
-------
A.3-3
to the surface of small particles and gases, followed,
respectively, by impaction and sorption or reaction. Also,
the" amount of material so deposited. (Units: ML~2T~1.)
Ensemble Mean; The average over a number of individual
model runs in which only one or a few adjustable parameters
are allowed to change.
Eulerian Model; A mathematical model in which computations
are made successively at fixed points in space (as opposed to
Lagrangian models where computations are made following an air
parcel). Computation points are usually arranged in a fixed
grid, and the model is also known as a grid model.
Flux; A physical quantity, the amount (mass) of material
passing through ,a unit area in a unit of time. (Units:
ML-2
-------
A.3-4
Isopleth; A line drawn on a field of values which joins
points of equal value in time or space.
Lagrangian Model; A mathematical model in which computations
are made successively in the same air parcel(s) as it moves
along a trajectory. Because this type of model is based on
following an air parcel, it is also known as a trajectory model.
Loading (atmospheric); The amount of a pollutant in the atmos-
phere expressed in mass or concentration units. (May also be
expressed on a per unit time and/or area basis.)
Loading Surface; A term used interchangeably with deposition.
LRTAP; The long-range transport of air pollutants refers to
the processes, collectively, by which pollutants are transported,
transformed and deposited, on a regional scale (of the order of
hundreds to thousands of km).
Mb (Millibar) Level; A surface of constant pressure in the
atmosphere, identified by the pressure expressed in mb.
(Common pressure levels used in air quality modeling are 925
and 850 mb levels.)
Mixing Height; The height above the earth's surface of a
boundary layer inversion which is usually the upper limit of
turbulent mixing activity, and which inhibits upward flux of
pollutant.
Model; A quantitative simulation of the behaviour of
a portion of the environment.
-------
A.3-5
Model Evaluation; A procedure by which the validity and sen-
sitivity of a model is assessed. Usually the validity is
ascertained by comparing model outputs with measurements,
and the sensitivity assessed through a series of model runs
in which input parameter values are altered in sequence, and
the results intercompared.
Model Intercomparison; A procedure of comparing the results
of several models which have been run on specified data bases
and with (usually) specified values of model parameters.
Model Resolution; The ability of a model to distinguish
(utilize) small spatial or temporal changes in input variables.
Model Sensitivity; A model characteristic which is described
by the response of an output parameter to a unit change in an
input variable or a model parameter.
Model Validation; The part of model evaluation in which modeled
results are compared with measured values.
Oxides of Nitrogen; This term usually denotes the sum of nitric
oxide (NO) and nitrogen dioxide (NC>2). Other forms are
nitrate (NOj), nitrous oxide (N20), and dinitrogen pentoxide
(N205).
Oxides of Sulfur; This term usually denotes sulfur dioxide
(S02). Other forms are sulfur trioxide (303) which is uncommon,
and sulfate (S04).
Parameterization; The representation of a physical, chemical
or other process by a convenient mathematical expression
containing quantities (parameters) for which measurements or
estimates are usually available.
-------
A.3-6
Receptor; An organism, ecosystem or object which is the
direct or indirect recipient of atmospheric deposition.
Scavenging; The processes by which materials are incorporated
into precipitation elements and (usually) brought to the earth's
surface.
Scenario; In the modeling context, a set of specified conditions
(usually emissions inventory) for input to the model which usually
reflect some anticipated future situation (e.g., energy use or
pollution emissions).
Sensitive Area; A geographical area in which a receptor (or
receptors) exhibit damage in response to a (pollution-imposed)
stress.
Sensitivity Receptor; The degree to which a receptor exhibits
an adverse effect from a (pollution-imposed) stress.
Source-Receptor Relationship; An expression of how a pollution-
source area and a receptor region are quantitatively linked.
Spatial Resolution; The minimum distance in space over which
meaningful differences in results can be determined (using a
particular model.) (For example, a model based on a 381-km
grid will provide no significantly different information for
two receptor points separated by less than approximately 381 km.)
Stationarity; Turbulent flow field is stationary when the flow
characteristics remain independent of the initial conditions.
Statistical Model; A mathematical model which uses statistical
values of parameters as inputs for the computations.
-------
A.3-7
Surrogate; The term applied to a parameter which is used to
represent another. (For example, modeling hydrogen ion
behavior in the atmosphere is difficult, so that sulfate ion
is used as a substitute.)
Susceptibility: A receptor or receptor area is said to be
susceptible if it is both sensitive, and receiving a pollutant
loading or stress.
Temporal Resolution; The minimum time during which meaningful
differences in results can be determined (using a particular
model). (For example, models using upper air data which are
only available every six hours are limited in their temporal
resolution to.about 6 hours.)
Trajectory; The path or track of an air parcel through the
atmosphere. It can be calculated from observed or gridded
wind data either forward or backward from a point (source or
receptor, respectively).
Transfer Matrix; A presentation of source-receptor relation-
ships in a matrix form. Matrix elements can be expressed
as percentage values, as absolute values, or as values
normalized by source strength.) Such a presentation provides
a means of easy comparison of the impact of a variety of
sources on a variety of receptors.
Transformation (chemical); The processes by which chemical
species are converted into other chemical species (in the
atmosphere).
-------
A.3-8
Variance; A measure of variability. It is denoted by o 2
and defined as the mean-square deviation from the mean, that
is, the mean of the squares of the differences between
individual values of x and the mean value 3c.
6"^ = E [(x-x)2], where E denotes the expected value.
Wet Deposition; Collectively, the processes by which materials
are removed from the atmosphere and deposited at the surface
of the earth by precipitation elements. The processes include
in-cloud and below-cloud scavenging of both gaseous and
particulate materials. Also, the amount of material so
deposited. (Units: ML~2T~1.)
-------
APPENDIX 4
Compilation of Attendance, Agenda and Minutes
for the Modeling Subgroup Workshops and
Work Group 2 Meetings
-------
A.4-1
Document 2-16 Available From:
James W.S. Young
Air Quality and Inter-Environmental
Research Branch
Atmospheric Environment Service
Environment Canada
4905 Dufferin Street
Downsview, Ontario M3H5T4
(416) 667-4786
Brand L. Niemann
Program Integration and Policy Staff (RD-681)
Office of Research and Development
Environmental Protection Agency
Washington, D.C. 20460
(202) 426-9434
-------
APPENDIX 5
List of Work Group 2 Reports
and Other Documents
-------
A. 5-1
List of Work Group 2 Reports and Other Documents
Revised
(July 3, 1981)
Identification
Phase
I
Document Name
WG2 Phase I Interim Report
Addendum to Appendix 6
Addendum to Appendix 8
Date
1/14/81
1/14/81
2/20/81
Number
2-1
2-2
2-3
II Unified S02 Emission Inventory
(1976-1980)
AES-LRT Model Profile
ASTRAP Model Profile
ENAMAP Model Profile
OME-LRT Model Profile
RCDM Model Profile
UMACID Model Profile
MEP-TRANS Model Profile
CAPITA-Monte Carlo Model Profile
6/30/81
5/15/81
5/12/81
6/30/81
3/31/81
7/10/81
6/24/81 (Interim)
6/30/81 (Interim)
6/30/81 (Interim)
Modeling Sub-Group Phase II
Report 7/10/81
Atmospheric Sciences Sub-Group
Phas<% II Report 7/10/81
Sulfur and Nitrogen Chemistry
in LRT Models
Trends in Precipitation
Composition and Deposition
Seasonal Dependence of Atmos-
pheric Deposition and Chemical
Transformation Rates for Sulfur
and Nitrogen Compounds
Global Distribution of Acidic
Precipitation and the
Implications for Eastern
North America
WG2 Phase II "Working Report"
Compilation of Attendance,
Agenda, and Minutes
7/10/81
7/10/81
2-4
2-5
2-6
2-7
2-8
2-9
2-10
2-11
2-12
2-13
2-14
2-14A
2-14B
2-14C
2-14D
2-15
2-16
-------
APPENDIX 6
A Simple Example of the Role of Models
in the Development of Emission Control Strategies
-------
A.6-1
A SIMPLE EXAMPLE OF THE ROLE OF MODELS IN THE DEVELOPMENT
OF EMISSION CONTROL STRATEGIES
A6.1 Introduction
This section outlines the use of transfer matrices
generated by the models for evaluating emission limits for
selected source regons given the desired air quality objectives.
Transfer matrices are based on the assumption that the average
concentration or deposition of a pollutant in any receptor
area is a linear combination of emissions of its precursor in
every source region. For example, deposition at receptor 1
is given by
+ T12E2 + Tl3E3 + ............ + T].NEN (A6.1)
where Ej is the emission rate in region j and T]_j are a set
of transfer coefficients. A similar relationship for deposition
holds for other receptors except that a different set of
coefficients T^j will apply. Thus a two-dimensional array of
transfer coefficients (a matrix) is required to describe the
deposition values at all receptors in terms of emissions from
all source regions.
Using matrix notation the relationship between sources
and deposition may be expressed as
If . T = D
S\s
or (A6.2)
N
^"E-jTi-; = Dif i = 1, ...... M
3=1
where N is the total number of source regions and M the total
number of receptors.
-------
A..6-2
The transfer matrix T is generated using a model/ or
models, of long-range atmospheric transport to simulate the
movement of pollutants emitted from sources in each emitter
9
region as they are dispersed across the receptor areas. Then
the matrix element T^j is calculated by separating out the
deposition produced at receptor i by emission in region j and
dividing by the magnitude of that emission. The transfer
matrix may then be combined with selected emission vectors E
representing various emission scenarios to assess the impact
on the deposition array D.
A6. 2 Illustration of Transfer Matrix Use in Assessment
To show the use of transfer matrices, a simple two source
and two receptor example is now presented. The problem to be
solved is - what are the allowable emissions for the source
regions which permit attainment to deposition limits at the
receptors and which meet the constraints on the magnitude of
the emissions?
The target depositions and the constraints on the
emissions for the two dimensional problem may be expressed
mathematically as
(A6.3)
Ei\ ^ / EMAXi
(A6.4)
where EMINj is the minimum emission region j is able to attain
-------
A.6-3
and EMAXj is the maximum emission region j is permitted. These
constraints define a set of feasible solutions. Graphically
these feasible solutions form an area in the two dimensional
variable space of (Ej_, £2). The boundaries of the areas are
defined by the six lines in (Ej_, £2) space defined in Equations
(A6.3) and (A6.4), namely:
E1T11 + E2T12 = Dl (line 1)
E1T12 + E2T22 = D2 (line 2)
(A5.5)
EI = EMIt^ (line 3)
£2 = EMIN2 (line 4)
EI = EMAX;L (line 5)
E2 = EMAX2 (line 6)
»
Figure A6-1 shows a schematic diagram of the region of feasible
solutions as defined by a set of lines such as those given in
lines 3 to 5.
Although the sequence of equations in lines 3 to 5
define the set of feasible solutions the selection of an
optimal solution requires a further criterion. This criterion
will define the objective of the controls strategy which is
usually the minimization of abatement costs while meeting
the constraints on deposition and emissions. For example,
assume that control costs are directly proportional to the
magnitude of the reduction in emission. Hence, to minimize
cost, the emissions must be maximized, i.e.
Maximize
subject to
constraint
= G(E) (line 7) (A6.6)
-------
A. 6-4
\
\
\
\
\
(line 3)
(line 2)
E]T2l
E2T22 =
(line 5)
\
optimal feasible
solution (E]_*, E2*)
EMAX2
EMIN2
infinite
t of
parallel lines
defined by
FEASIBLE^SOLUT ION
(line 6)
E2 = EMAX2
(line 1)
E1T11+E2T12=D1
S j E = csDnstant
(line 4)
E2 = EMIN2
EMAXi
Figure A6-1:
Graphic representation of control strategy assessment,
Lines 1-6 define the two-dimensional solution space
of the control strategy problem. The diagonal solid
lines denote solutions to the objective function for
the control scenario and the dot marks the optimal
solution (E]_*, E2*) for the example.
-------
A.6-5
Equation (A6.6) is the objective function which is used
to select the optimal solution. The relationship
N,
"^_I E-; = constant
j=l
t
defines an infinite set of parallel lines through the region
of feasible solutions for which G(E) is the constant describing
the line intersecting the region at the optimal solution (or
solution). The values of Ej_ and £2 at the point marked by
the dot in Figure A6-1 are solution to this control strategy
exmaple. It is evident from this two-dimensional example
that the line passing through the dot has the maximum intercepts
on EI and £2 axes, of all lines parallel to it and having a
segment in the shaded region. This establishes the optimality
of .the feasible solution.
-------